MRI vs. Ultrasound for DVT
Abstract & Commentary
Synopsis: The sensitivity and specificity of MRI were similar to venography and ultrasound.
Source: Fraser DG, et al. Ann Intern Med. 2002;136:89-98.
This trial evaluated the efficacy of MRI in diagnosing DVT. The trial enrolled 338 consecutive patients suspected of having a DVT. Venography was successfully performed in 298 of these patients; this was used as the gold standard for the presence or absence of DVT. All patients with a positive venography (n = 58) and 48 patients with a negative venography were included in the study. By study protocol, every fourth patient with a negative venography was included in the study.
Selected patients then underwent MRI of both lower extremities. Two reviewers blinded to the venography results, an expert radiologist and a non-radiologist trained in MRI, interpreted the MRIs. The reviewers were asked to comment on the presence or absence of DVT, and the location of the DVT (isolated calf, femoropopliteal, and ileofemoral).
The overall sensitivity of MRI was 96% (95% CI, 89-99%) and the specificity was 90% (95% CI, 79-96%). The MRI was 100% sensitive and specific for ileofemoral DVTs, and 97% sensitive and 100% specific for femoropoliteal DVTs. The MRI was less sensitive (reviewer 1: 92%; reviewer 2: 83%) and specific (reviewer 1: 94%; reviewer 2: 96%) for isolated calf DVTs.
There was good intra-observer agreement (k = 0.94; 95% CI, 0.88-1.0)
Comment by Jeff Wiese, MD
Venography, the gold standard for diagnosing DVT, has largely been replaced by noninvasive tests such as ultrasound or impedence plethysmography. Ultrasound is 89% sensitive and 97% specific for DVTs above the knee, but is limited in its ability to diagnose pelvic DVTs and DVT recurrence.1 Other limitations of venography and ultrasound include its use in patients with full-length leg casts, patients suspected of having pelvic DVTs, or patients suspected of having DVT recurrence.
This study confirms that MRI is a viable option for diagnosing DVT. The positive and negative likelihood ratios for MRI are 9.6 and 0.04.2 This study is limited by the high number (26%) of participants that dropped out after enrollment. It does not appear, however, that the loss of these patients from the protocol would have changed the results. The sensitivity and specificity of MRI were similar to venography and ultrasound.
The primary limitation of MRI is the cost and access to the test. The MRI protocol in this study used a standard clinical magnet with no special patient preparation or use of contrast. The cost of the MRI depends on the scanning time. On average, a 40-minute scanning time costs $1000. The average scanning time in this study was 12 minutes, making the estimated cost of MRI of $300 comparable to ultrasound ($300) and venography ($450). The scanning time depends on the experience of the technician, the skill of the radiologist, and the access to the MRI scanner. All of these parameters will vary from hospital to hospital; clinicians should make the decision as to the use of MRI in diagnosing DVT based on their access to MRI scanning, the facilities of their individual hospital, and the scanning time used in their hospital.
Understanding the physiology of MRI is important to selecting patient subgroups that may benefit from MRI. Methemoglobin production increases once blood clots, and this shortens the T1 magnetic resonance. The result is an increased density seen within the vein on MRI. The production of methemoglobin declines over time, enabling MRI to distinguish acute (high density) from chronic (low density) DVTs.3 Unlike ultrasound or venography, MRI does not depend on venous occlusion of flow for diagnosis. This makes MRI less likely to be falsely negative in patients with venous underfilling due to poor circulation or dehydration.
Patients that may particularly benefit from MRI include pregnant patients (due to MRI’s ability to diagnose pelvic DVTs), patients with full-length leg casts, and those with a history of chronic DVTs being evaluated for DVT recurrence.4
Measurement of ELISA d-dimer levels was not performed in the study, and the addition of this test to diagnostic protocols may improve the specificity of the diagnostic protocol.
Dr. Wiese, Chief of Medicine, Charity, and University Hospitals, Associate Chairman of Medicine, Tulane Health Sciences Center, is Associate Editor of Internal Medicine Alert.