Detection of pulmonary embolism using repeated MRI acquisitions without respiratory gating: a preliminary study

Abstract

Background

Pulmonary embolism (PE) is a severe medical condition with non-specific clinical findings. Computed tomography angiography (CTA) using iodinated contrast agents is the golden standard for diagnosis, but many patients have contraindications for CTA.

Purpose

To investigate the diagnostic accuracy of repeated acquisitions of magnetic resonance imaging (MRI), without respiratory gating or breath holding, in diagnosing PE using CTA as the reference standard.

Material and Methods

Thirty-three patients with clinically suspected PE underwent MRI within 48 h after diagnostic CTA. A control group of 37 healthy participants underwent MRI and was matched with an equal number of negative CTA exams. The MRI protocol was based on free-breathing steady-state free precession producing 4.5 mm slices in axial, sagittal, and coronal planes. Instead of respiratory or cardiac gating five repetitive slices were obtained in each anatomical position to compensate for movement and artifacts. Clinical assessment including d-dimer and Well’s score was performed prior to imaging. One radiologist reviewed the CTA exams and two radiologists reviewed the MRI scans.

Results

All 70 MRI exams were of diagnostic quality and the total acquisition time for each MRI scan was 9 min 34 s. On CTA, 29 patients were diagnosed with PE and the MRI readers detected 26 and 27 of those, respectively. Specificity was 100% for both readers. Sensitivity was 90% and 93%, respectively. Inter-reader agreement using Cohen’s kappa was 0.97.

Conclusion

Our unenhanced MRI protocol shows a high sensitivity and specificity for PE, but further studies are required before considering it as a safe diagnostic test.

Keywords

Pulmonary embolism thromboembolic lung steady-state free precession (SSFP)

Introduction

Pulmonary embolism (PE) is the third most common acute cause of cardiovascular death (1,2). Due to non-specific clinical findings, diagnostic imaging has become central for the diagnosis of PE (3,4). At present, computed tomography angiography (CTA) has become the modality of choice for diagnosing PE in clinical practice, due to its high sensitivity and specificity in addition to extensive availability and short acquisition time (1,3,5). Limitations regarding the use of CTA include exposure to ionizing radiation and the use of iodinated contrast material (2,4,6). It has been estimated that about 25% of patients with suspected PE have contraindications for CTA, such as renal insufficiency (7). However, MRI is not yet validated as a second line test for PE diagnosis in patients with contraindications for CTA (8,9) although recent technical improvements in MRI suggest it as a possible alternative (6,10 –12).

Different MRI protocols, of varying diagnostic accuracy, have been investigated to detect PE (6), the three most commonly used being gadolinium-enhanced magnetic resonance angiography (Gd-MRA), unenhanced MRI sequences, and magnetic resonance perfusion (8,13). The largest study so far (PIOPED III) was a prospective multicenter study, investigating the diagnostic accuracy of gadolinium-enhanced magnetic resonance pulmonary angiography and thigh venography in acute PE. It concluded that technically inadequate investigations varied among hospitals from 11–52% and therefore only should be considered at hospitals that routinely perform it well and only in patients for whom CTA is contraindicated (14). Among the studies on unenhanced MRI to detect PE, the sequences generally have shown a lower sensitivity compared to other MRI protocols, but a high specificity (98–100%), suggesting that a positive finding is sufficient to initiate anticoagulation therapy (4,8).

We hypothesized that an unenhanced MRI protocol repeating multiple (five) slices in each anatomical position in three orthogonal planes without any cardiac or respiratory gating could be used to detect PE. The purpose of the present study was to investigate the diagnostic accuracy of such an unenhanced MRI protocol.

Material and Methods

Patients

The study was performed according to the Declaration of Helsinki and approved by the local Ethics Committee. Written informed consent was obtained from each patient.

Patients with clinically suspected pulmonary embolism that had undergone diagnostic CTA were selected for the study. Exclusion criteria were contraindications to MRI and more than 48 h between CTA and MRI exams.

From February 2012 to January 2014, 70 participants were included in the study. All patients were included by the same two clinicians, therefore patients could only be included when they were on duty. Due to the time between CTA and MRI, mainly patients admitted to the hospital could be included. One patient was excluded due to a MRI-incompatible breast implant (a 51-year-old woman). A group of 33 patients (23 men, 10 women; average age, 48 years; age range, 22–87 years) with clinically suspected PE underwent MRI within 48 h of diagnostic CTA. There was no delay in anticoagulation treatment in patients diagnosed with PE prior to MRI. Due to a small number of patients with a negative CTA (2 women; ages, 45 and 74 years; and 2 men; ages, 64 and 75 years), a control group of 37 healthy participants (9 men, 28 women; average age, 48 years; age range, 26–66 years) underwent MRI and was matched with an equal number of consecutive negative CTA exams from the hospital patient flow (15 men, 22 women; average age, 57 years; age range, 23–83 years).

The MRI and CTA exams from the first cohort of 33 participants and the control examinations were mixed and blinded before image analysis.

Clinical assessment

A clinical assessment was made at the emergency care unit and recorded on a standardized form. Symptoms (cough, dyspnea, chest pain, syncope), pulse, blood pressure, oxygen saturation, temperature, electrocardiography, and laboratory findings including d-dimer and Well’s score were recorded prior to imaging.

CTA

All CTA exams were performed by a 64-section CT scanner, Lightspeed VCT, GE Healthcare, Milwaukee, WI, USA). The amount of contrast agent (Ultravist 370 mg I/mL, Bayer HealthCare Pharmaceuticals Inc., Berlin, Germany) and injection rate was determined by the patients’ age, weight, and height according to a standardized protocol. Parameters: 100 kV; automated tube current modified by a dose regulation program; noise index, 40; pitch, 1.37:1; rotation time, 0.4 s; coverage per rotation, 40 mm; slice thickness, 0.625 mm; reconstruction slice thickness, 2 mm in three orthogonal planes.

MRI

Patients were placed (supine, head-first) in a 1.5 T MRI scanner (Magnetom Aera, Siemens Medical Systems, Erlangen, Germany). A dorsal 32-element spine matrix coil integrated in the table and a ventral 18-element body matrix coil were used. The MRI protocol was based on a 2D free-breathing steady-state free precession (SSFP) sequences, without use of any intravenous contrast agent, respiratory, or electrocardiographic gating. The patients received no specific breathing instructions.

Parameters: flip angle, 70°; field of view, 450 mm; matrix size, 256 × 256; voxel size, 0.9 × 0.9 × 4.5 mm; TE, 1.23–1.26 ms; TR, 2.8 ms; slice thickness, 4.5 mm; overlap, −2.7 mm (60%); acquisition time, 9 min 34 s. Acquisition time in each plane: axial, 3 min 50 s; sagittal and coronal, 2 min 52 s each).

Five slices were obtained in each anatomical position in three orthogonal planes, generating stacks of approximately 500 images in each plane (axial, 600; sagittal, 450; and coronal, 450). The 1500 images were sorted by position, generating stacks with multiple images in various phases of the breathing and cardiac cycle.

Image analysis

Before analysis, all patient data were removed from the images. The exams from the patient group and the control group were randomly mixed. The presence of PE was based exclusively on vascular signs such as a complete filling defect, a partial filling defect and/or a “railway track” sign (freely floating thromboembolic masses in the vessel) (2,15). Secondary signs were not used to imply or exclude embolus.

The MRI exams were reviewed by two radiologists (R1 and R2) blinded to the CTA exams; R1 had 1 year of sub-specialty experience in thoracic radiology and cardiac MRI. R2 had 1 year of sub-specialization in thoracic radiology. The MRI readers obtained all the five slices from each position sorted in stacks by anatomical position. The CTA exams were reviewed by a thoracic radiologist with more than 10 years of experience of thoracic radiology. The CTA interpreter was blinded to the MRI exams. After recording each reader’s individual results, a consensus reading was also performed in cases of disagreement between the methods, and reported separately.

The pulmonary vascular bed was divided into vascular territories according to the system used by Kalb (5): right pulmonary artery, right upper lobe, middle lobe, right lower lobe, left pulmonary artery, left upper lobe, lingual, and left lower lobe, thus assigning the segmental and subsegmental arteries to their supplying lobar artery. If an embolus was detected, the vasculature distal to it was not studied further.

Statistical analysis

MedCalc software (16) was used to calculate sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) with 95% confidence intervals (CI). Kappa values were also calculated to determine inter-reader agreement using GraphPad software (17).

Results

Regarding clinical data, d-dimer was obtained in 24 participants. D-dimer was positive in 23 patients and negative in one patient, both with a fixed cutoff at 500 µg/L and with age-adjustment according to Righini et al. (18). PE was considered unlikely if Well’s score ≤ 4 points and likely if > 4 points, resulting in 16 patients with likely PE and 11 unlikely. In two patients Well’s score was not recorded. Details regarding blood pressure, heart frequency and oxygen saturation are given in Table 1.

Table 1.

Summary of the clinical data of patients with CTA-confirmed PE.

Age	Gender	D-dimer	Wells score	Symptoms	Heart rate (bpm)	Blood pressure (mmHg)	SaO2 (%)	Level of PE
87	M	N/A	Likely	Dyspnea	110	140/70	95	Lobar
83	F	Pos	Likely	Cough, dyspnea, CP	115	200/94	94	Lobar
75	M	Pos	Unlikely	Cough, dyspnea, CP	95	135/80	92	Lobar
74	M	Pos	Unlikely	Dyspnea	79	145/80	94	Lobar
70	M	Pos	Likely	Dyspnea, CP, syncope	77	160/74	97	Central
68	M	Pos	Likely	Cough, dyspnea, CP	113	106/78	83	Central
66	M	N/A	Likely	Dyspnea, CP	72	240/120	96	Central
63	F	Pos	Likely	None	76	120/60	95	Central
30	M	Pos	Likely	Cough, dyspnea	106	176/96	98	Lobar
56	M	Pos	Likely	Dyspnea	140	135/75	94	Central
56	M	Pos	Likely	Cough, dyspnea, CP	86	160/100	97	Lobar
55	F	Pos	Likely	Cough, dyspnea	117	138/74	94	Central
53	M	Pos	Likely	Dyspnea, CP	80	121/75	100	Segmental
51	M	Pos	Unlikely	Cough	66	160/75	96	Segmental
51	F	Pos	Likely	Cough, dyspnea, presyncope	130	137/82	88	Central
52	M	Pos	Unlikely	Cough, dyspnea	90	140/80	97	Central
52	M	Pos	Unlikely	Dyspnea	87	160/100	90	Central
50	M	Pos	Likely	Cough, dyspnea	74	117/74	97	Lobar
49	M	N/A	N/A	N/A	73	135/77	95	Segmental
49*	M	N/A	Likely	None	61	140/86	99	Subsegmental
46	M	Pos	Unlikely	Cough, dyspnea, CP	95	150/110	88	Central
42	M	Pos	Unlikely	CP	80	132/83	97	Segmental
40	F	Pos	Unlikely	CP	98	135/80	99	Subsegmental
36*	F	Neg	Unlikely	Dyspnea	65	110/80	100	Segmental
34	M	Pos	Likley	Cough, dyspnea	100	125/76	95	Central
22	M	Pos	Likely	Cough, dyspnea, CP	66	134/73	98	Central
22	F	Pos	Unlikely	Cough, dyspnea	74	125/71	100	Central
50	F	Pos	Unlikely	Dyspnea, syncope	110	103/65	91	Central
58	M	N/A	N/A	CP	83	160/100	91	Subsegmental
	F = 8	Pos: 23	L = 16
54	M = 21	Neg: 1	U = 11		90		95

Age-adjusted d-dimer defined as positive (pos) or negative (neg). Wells score was defined as unlikely (U) if ≤ 4 point and likely (L) if > 4 points. Symptoms recorded were dyspnea, cough, chest pain (CP), and syncope. Level of PE refers to level of the most proximal embolus detected.

The two patients with false-negative MRI.

SaO2, oxygen saturation measured by pulsoximeter.

Data from 70 CTA and MRI examinations was obtained. All of the CTA and MRI exams were complete and of diagnostic quality. For the cohort that underwent both CTA and MRI (n = 33), the mean time between CTA and subsequent MRI was 22 h 39 min (time span, 4 h 28 min to 47 h 22 min). The total acquisition time for the MRI scan was 9 min 34 s.

On CTA, 29 patients were diagnosed with PE, proximal in 21 patients and limited to the segmental or subsegmental level in eight patients (Table 1). On MRI, 27 and 26 of the 29 patients were detected by R1 and R2, respectively. Sensitivity for R1 with the MRI protocol used was 93% (95% CI, 76–99%) and for R2 90% (95% CI, 73–98). Specificity was 100% (95% CI, 91–100%) for both readers. PPV was 100% for both readers (95% CI, 87–100%). NPV was 95% (95% CI, 84–99%) for R1 and 93% (95% CI, 81–99%) for R2. Inter-reader agreement between R1 and R2 was very good, with a kappa value of 0.97 (95% CI, 0.91–1.00).

After the individual readings had taken place, a consensus reading was made by the CTA reader and the MRI readers. It was agreed that isolated subsegmental PE was found on the CTA exams but not on the MRIs in the two patients that were described as false negative by both MRI readers. It was also found that several emboli had moved more peripherally or possibly disappeared in several patients between the CTA and MRI exams.

Four patients in the study, only had emboli in one or two sub-segmental branches and it was in two of those patients that PE was not detected on MRI. In the two patients (ages 36 and 41 years) missed on MRI time between CTA and MRI was 16 h 58 min and 39 h 16 min, respectively. The d-dimer was 390 and 900 µg/L and Well’s score 0 and 3, respectively. In the patients (ages 51and 58 years) with a single subsegmental embolus detected on both CTA and MRI examinations, the time between the investigations was 17 h 2 min and 22 h 45 min, respectively. The d-dimer 1140 µg/L with and Well’s score 1 in one patient and not available in the other. Image examples of MRI and CTA in one of the false-negative and true-positive cases with subsegmental emboli are given in Figs. 1 and 2.

Fig. 1.

(a) CTA from a 52-year-old man with an isolated subsegmental embolus (arrow) in the right lower lobe. (b) Corresponding MRI in the same patient 17 h 2 min after CTA shows the isolated subsegmental embolus (arrow).

Fig. 2.

(a) CTA from a 41-year-old woman with an isolated subsegmental embolus (arrow) in the right lower lobe. (b) Corresponding MRI in the same patient 39 h 16 min after CTA. The vessel is identified (arrow) but shows no remaining embolus.

Discussion

The principal finding in our study is that unenhanced MRI with repeated acquisitions has a high sensitivity and specificity in diagnosing PE. In our protocol, five repeated acquisitions in each anatomical position was acquired. No breath-holding was required, still no cardiac or respiratory gating was used. This method, not previously reported in the literature, produces non-gated images in various positions and flow conditions in the pulmonary arteries, which possibly increases the ability to discriminate emboli from artifact.

The specificity for the diagnosis of PE with our MRI protocol was 100%, which is consistent with previous studies (4,5,8). The MRI readers identified 26 or 27, respectively, out of 29 patients with PE on CTA (sensitivity, 90–93%), which is also similar to previous studies on unenhanced MRI (4,5,8), even though the radiologists had very limited experience with this method. For both readers, two of the false-negative MRI exams consisted of two patients with isolated subsegmental emboli. Comparing them with the only patient in the study with isolated subsegmental PE on both CTA and MRI, it was observed that the time between investigations was longer and the patients younger in the false-negative cases.

The prevalence of isolated subsegmental emboli is in the range of 7–36% among patients with PE (11). In our study, the prevalence was 13%. The significance of subsegmental PE has been questioned, in particular taking into consideration the risk of anticoagulation therapy (18). However, it has been proposed that they might predict a potential risk for future embolism and pulmonary hypertension (11). Although CTA seems to allow improved visualization of subsegmental arteries, its accuracy is difficult to determine; for instance, reader disagreement has been estimated at 30–37% in different studies (18).

There have been several different MRI protocols used to diagnose PE in different study protocols, Gd-MRA, real-time MRI, and MR perfusion being the three most frequently used (11). Oudekerk et al. proposed that MRA is both sensitive and specific for segmental or larger emboli, with a sensitivity for diagnosing patients with PE of 77% and specificity of 95% (19). In central and lobar arteries, the sensitivity was 100%, in the segmental 84% and in the sub-segmental only 40% (19). Regarding SSFP, a study by Kalb et al. consisting of 22 patients showed a specificity of 100% and sensitivity of 67% for respiratory triggered SSFP (4), while a study on 274 patients by Revel et al. showed a specificity of 90% and sensitivity of 82% for unenhanced free-breathing SSFP (8). Kalb et al. also showed one of the potential hazards of respiratory gating, where 1/22 patents had a poor respiratory triggering rendering in a technically inadequate investigation (4) whereas our MRI protocol is less susceptible to similar technical problems. Our protocol also renders several captions of the same location making it easier for the reader to identify potential artifacts (Fig. 3).

Fig. 3.

(a–e) Show the repetitive five MRI-slices in one anatomical position of a 51-year-old woman with emboli in the right pulmonary artery and bilateral in upper lobe arteries.

Unlike several previous studies using different MRI sequences for diagnosing PE (4,5,8,14), no technically inadequate MRI exams occurred in our study. Therefore, it may be argued that our protocol is easily applied for radiology technicians. The total acquisition time of the MRI exam was less than 10 min, making it possible to use in a clinical setting for emergency medicine. Furthermore, patients unable of holding their breath may be successfully investigated.

There are a number of limitations in our study. First, the recruitment of patients: most of the participants were already admitted to the hospital at the time of inclusion in the study and may be sicker than an unselected population. That is probably why a large proportion of the participants (29 out of 33) had PE; this is not representative of the overall occurrence of PE in general, which is approximately 20% (20,21). Second, the size of the study was small, with only 70 participants. Third, the time between CTA and MRI: the mean time between the investigations was 22 h 45 min. During this time, the patients diagnosed with PE received anticoagulation therapy, which might have resolved the subsegmental emboli between the two exams, resulting in a false, low sensitivity. Also it appeared that several emboli had moved more peripherally, probably due to treatment. To improve the patient recruitment and reduce the effect of treatment MRI should preferably be performed directly before or after CTA.

In conclusion, our results suggest that unenhanced MRI according to the protocol used in the future may be considered for patients where CTA is contraindicated. However, due to the relatively small size of the present study further clinical validation is required, particularly since the MRI protocol used has not been investigated before.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Swedish Heart and Lung Foundation, the Swedish Society of Medicine, and Stockholm City Council.

References

Goldhaber

Bounameaux

. Pulmonary embolism and deep vein thrombosis. Lancet 2012; 12: 1835–1846.

Hochhegger

Ley-Zaporozhan

Marchiori

, et al. Magnetic resonance imaging findings in acute pulmonary embolism. Br J Radiol 2011; 84: 282–287.

Bounameaux

Perrier

Righini

. Diagnosis of venous thromboembolism: an update. Vasc Med 2010; 15: 399–406.

Kalb

Sharma

Tigges

, et al. MR imaging of pulmonary embolism: diagnostic accuracy of contrast-enhanced 3D MR pulmonary angiography, contrast-enhanced low-flip angle 3D GRE, and nonenhanced free-induction FISP sequences. Radiology 2012; 263: 271–278.

Kluge

Mueller

Strunk

, et al. Experience in 207 combined MRI examinations for acute pulmonary embolism and deep vein thrombosis. Am J Roentgenol 2006; 186: 1686–1696.

Zhou

Long

, et al. Diagnostic performance of magnetic resonance imaging for acute pulmonary embolism: a systematic review and meta-analysis. J Thromb Haemost 2015; 13: 1623–1634.

Stein

Fowler

Goodman

, et al. Multidetector computed tomography for acute pulmonary embolism. N Engl J Med 2006; 354: 2317–2327.

Revel

Sanchez

Lefort

, et al. Diagnostic accuracy of unenhanced, contrast-enhanced perfusion and angiographic MRI sequences for pulmonary embolism diagnosis: results of independent sequence readings. Eur Radiol 2013; 23: 2374–2382.

Davidson

Lacrampe

. Why can’t magnetic resonance imaging reliably diagnose pulmonary embolism? Ann Intern Med 2010; 152: 467–468.

10.

Hochhegger

Marchiori

Irion

, et al. Magnetic resonance of the lung: a step forward in the study of lung disease. J Bras Pneumol 2012; 38: 105–115.

11.

Clemens

Leeper

Jr . Newer modalities for detection of pulmonary emboli. Am J Med 2007; 120: S2–S12.

12.

Henzler

Schmid-Bindert

Schoenberg

, et al. Diffusion and perfusion MRI of the lung and mediastinum. Eur J Radiol 2010; 76: 329–336.

13.

Francois

Hartung

Reeder

, et al. MRI for acute chest pain: current state of the art. J Magn Reson Imaging 2013; 37: 1290–1300.

14.

Stein

Chenevert

Fowler

, et al. Gadolinium-enhanced magnetic resonance angiography for pulmonary embolism: a multicenter prospective study (PIOPED III). Ann Intern Med 2010; 152: 434–443.

15.

Ohno

Higashino

Takenaka

, et al. MR angiography with sensitivity encoding (SENSE) for suspected pulmonary embolism: comparison with MDCT and ventilation-perfusion scintigraphy. Am J Roentgenol 2004; 183: 91–98.

16.

MedCalc. MedCalc easy-to-use statistical software. Ostend: MedCalc Software bvba. Available at: https://www.medcalc.org (accessed 11 November 2015).

17.

Software G. GraphPad Software. 2015.

18.

Righini

Van Es

Den Exter

, et al. Age-adjusted d-dimer cutoff levels to rule out pulmonary embolism. JAMA 2014; 311: 1117–1124.

19.

Goodman

. Small pulmonary emboli: what do we know? Radiology 2005; 234: 654–658.

20.

Oudkerk

van Beek

Wielopolski

, et al. Comparison of contrast-enhanced magnetic resonance angiography and conventional pulmonary angiography for the diagnosis of pulmonary embolism: a prospective study. Lancet 2002; 359: 1643–1647.

21.

Agnelli

Becattini

. Acute pulmonary embolism. N Engl JMed 2010; 363: 266–274.