To better understand the technical and logistical details of performing pulmonary MRA (pMRA) for detection of pulmonary embolism (PE) in Emergency Room (ER) patients. Specifically, the attendee will learn when and why pMRA is a viable option for ruling out PE, the pulse sequences and techniques used for performing pMRA, and some fundamentals of pMRA interpretation. Furthermore, this course will discuss how one might go about introducing pMRA to ER physicians and adding pMRA to the MR service.

Pulmonary embolism is not uncommon, with an incidence of 4-21/10,000, rising to on the order of 5 times this in patients greater than age 80. Left undiagnosed and untreated, mortality can be greater than 50% [1]. Frustratingly to our clinician colleagues, PE is difficult to diagnose and presents in many different ways. In testament to this, despite clinical vigilance and advanced medical imaging techniques, autopsy studies reveal that only approximately 50% of patients who die of PE are suspected of this condition pre-mortem [1].

Since PIOPED II [2], multi-detector CT pulmonary angiography (CTPA) has become the mainstay for the diagnosis of pulmonary embolism. The technique is fast, widely available, has essentially 100% positive predictive value (PPV), and its negative predictive value (NPV - based on clinical follow-up of patients having a “negative” CTPA) is extremely high, approaching 99% [3]. Nonetheless, CTPA is imperfect, known to miss up to 50% of isolated subsegmental PE’s [4]. Interestingly, while the detection of PE has convincingly increased since the advent of multi-detector CTPA, there was no apparent improvement in outcomes, suggesting that these missed/occult subsegmental isolated emboli may not be clinically significant [5].

Given how good a “gold standard” CTPA has become, it does have notable drawbacks. First, despite newer radiation dose reduction techniques, the radiation dose (6-10+ mSv) is not inconsequential, particularly in young patients, and breast doses can be much higher. Woo et al. estimate that in patients age 15-40, the lifetime attributable risk for cancer mortality due to a single CTPA ranges from 35-57/100,000, being highest in the young, particularly women [6]. Weigh this against the fact that in the 18-45 age range, only about 5% of CTPA’s are positive, meaning 20 young patients must be irradiated to diagnose a single PE [7]. Second, iodine can and does cause contrast nephropathy post CTPA [8]. Finally, many patients have iodine allergies making CTPA unsuitable.

MR angiography (MRA) has been around in various forms for more than 25 years, and has become a relatively mature technology that is used as a first-line imaging modality at many institutions for multiple vascular territories. Until recently, however, pulmonary MRA (pMRA) has faced challenges. In 2010, PIOPED III compared contrast-enhanced MRA (CE-MRA) to CTPA [9]. In this large, multicenter trial (n=371; 7 sites), 25% of the studies were non-diagnostic, and sensitivity was only 72% (although 99% specificity was observed). The trial concluded that pMRA should only be performed in centers that “perform it well”. While this dealt pulmonary MRA a setback, realize the imaging technology is 10 years old (2006 – 2008), and non-standardized, often sub-optimal techniques were employed. Since then, there have been considerable hardware advances, as well as better understanding of contrast injection strategies to decrease artifacts and how to more accurately interpret clinical images [10]. In recent work from the Wisconsin group, of 190 contrast-enhanced pMRA exams, 97% were diagnostic [11]. Following 167 “negative” patients there was a 97% NPV at 3 months, dropping slightly to 96% at 1 year; comparing favorably to similar CTPA studies. The more sophisticated algorithm employed by this group can be done very quickly (<10 minutes table time) and easily integrated into ER clinical workflow. Recent work by other investigators have demonstrated similar results [12,13]. These advances suggest that for appropriately triaged patients, namely those who can tolerate a breath-hold time of ~16-18 seconds, who are less than age 40 (particularly female), or who have contraindications to iodine contrast, optimally performed pMRA has high accuracy and eliminates potentially harmful effects of radiation exposure, especially considering the low (5%) prevalence of PE in this demographic.

Pulmonary MRA can be performed with or without contrast, although generally contrast-enhanced techniques are simpler, faster and more robust. The most commonly used pulse sequence is a post-contrast 3D spoiled gradient echo (T1-weighted), typically performed in the coronal plane using relatively high parallel imaging factors to achieve high spatial resolution (on the order of 1.1-1.3 x 1.3-1.8 x 2.0-2.6 mm³, interpolated to near 1 x 1 x 1 mm³) in a 12-20 second breath-hold (Figure 1) [11-13]. Timing can be performed using a small timing bolus, or by automated detection over the right ventricle or pulmonary trunk. Single dose gadolinium contrast is generally adequate, and it has been recently proven advantageous to dilute contrast such that it can be administered at a relatively high flow rate (1.5 – 2.0 mL/s) for a duration approximating the scan time (Maki unpublished data) [11]. This minimizes blurring and ringing that can occur if the contrast concentration varies over the acquisition time. Breath-holding can be inspiratory or expiratory. It is also important to obtain a second acquisition after allowing the patient to catch their breath for 10-15 sec. Contrast opacification is still quite adequate in this early delay, and this gives not only a second chance to obtain a diagnostic image (e.g. if timing incorrect or respiratory motion), but often can more effectively demonstrate intraluminal clot, as there is background perfusion so it is less seeing “black on black”. Furthermore, these two phases can give some indication of perfusion, where defects can help localize and confirm pulmonary emboli. Unlike CTPA, in cases of failure, repeat scanning with a second dose of contrast can be performed with no radiation penalty.

Diagnosis is typically made from native source images combined with coronal, sagittal and axial multiplanar (MPR) or thin overlapping maximum intensity projections (MIPs). Some authors suggest that additional pMRA sequences be performed, and Kalb et al. demonstrated significantly improved sensitivity when combining CE-MRA with pre-contrast ECG-triggered SSFP and post-contrast low flip angle gradient echo breath-hold imaging [12]. Notably, non-contrast techniques are a viable option for pregnant patients suspected of PE, where CTPA, V/Q scanning, and gadolinium administration are undesirable options.

Once gadolinium has been administered, vascular enhancement persists in excess of 10 minutes, and rapid MR venography (MRV) through the pelvis and lower extremities can, if desired, be performed using a low flip-angle, fat suppressed gradient echo sequence. Since venous thromboembolism is a necessary requisite for PE, and both are typically treated similarly, detecting significant venous thrombus can appropriately direct therapy even should pMRA be non-diagnostic. In the PIOPED III trial, the addition of MRV improved sensitivity from 78-99% while minimally decreasing specificity from 99-96% [9]

Offering pMRA to the ER requires timely access to a high-performance 1.5 or 3T MR unit, expertise (physician and technologist) to appropriately utilize the MRA equipment, experienced interpreters, careful planning and patient triage, and buy in from multiple parties; ER, MR service, and hospital administration. The argument in favor can be based primarily on dose/risk reduction in the young population. This must be carefully explained to the ER physicians based on our best understanding of radiation risk and our duty to the principle of ALARA (keeping radiation dose As Low As Reasonably Achievable). Concerns often arise such as “what if we miss a PE?”, “what about other diagnoses that may be uncovered by CTPA?”, or “we can’t afford to delay as long as it often takes to get a MRI”.

The answer to the first question has been covered above as reflected in studies such as Wiener and Schiebler [5,11], and the desire to not “overtreat”. In terms of what pMRA may “miss”, a study by Chandra et al. determined CTPA only found alternative diagnoses not known or discovered on chest x-ray in 8% of patients [14]. Add to this that pMRA is excellent for detecting other vascular abnormalities such as dissection and aneurysm, and depending on how many sequence types are implemented, quite good for mediastinal mass, pleural effusion, non-trivial airspace abnormalities, rib fractures, and parenchymal abnormalities of the upper abdomen. Personal discussions with colleagues at University of Wisconsin and University of Arizona (institutions where large volume pMRA is performed) assure me that missing “other” significant abnormalities is rare. This does, however, speak to the need for instituting appropriate triage guidelines for your institution. Obviously patients unable to cooperate (e.g. claustrophobia) or hold their breath the requisite approximately 15 sec are better served with CTPA. Depending on the institution’s protocol and level of comfort, patients with known thoracic malignancies, infections, or other complex/confusing thoracic pathologies may not be optimal pMRA candidates. Addressing the final concern, expediency is important for ER patients. This requires buy in from the MR department and hospital administration to move these patients into the MRI quickly, most often “squeezed in” between other patients. This is simplified when the protocol is kept short, as per Schiebler et al. [11].

The bottom line is that implementing pMRA in the ER requires a discussion between all involved parties to determine how to best proceed, and how to select the most suitable patients. One tactic is a “staged” approach, where initially studies are performed using a more “comprehensive” protocol (e.g. CE-MRA plus ECG-triggered non-contrast MRA and/or delayed post-contrast GRE imaging and/or MRV) only on optimal patients during the daytime when the most highly skilled personnel are readily available. This allows the team to gather experience with some redundancy in the most optimal setting. During this initial period, careful ongoing evaluation by the radiologists and ER physicians must occur, and based on this protocol, logistical and patient-selection adjustments can be made. Once a level of comfort is achieved (radiologists, technologists, ER physicians), sequences thought to not add good value can be dropped (saving time and making it faster and easier to get patients into the scanner), the patient selection criteria can be expanded, and the hours of operation can be expanded. This latter step, at least in teaching hospitals, requires training of radiology trainees in pMRA interpretation, which in most cases is quite straight-forward.