Patient and lesion demographics
From the 18 patients included,
14 were female and 4 were male,
with a mean age of 56.2 ± 13.4 years.
5 lesions were malignant (2 metastasis,
lymphoma and sarcoma) and 13 were benign (6 myxomas,
2 mitral valve casseous necrosis,
2 pericardical cysts,
1 intracavitary thrombus and 1 hydatid cyst).
Attending to their composition,
4 lesions were cystic and 14 were solid.
Regarding its behavior after gadolinium based contrast,
38.9 % (n=7) of the tumors did not show enhancement.
From those lesions (61.1%; n=11) that showed enhancement at LGE did show enhancement,
in up to 91% was heterogeneous.
The majority of the lesions were hyperintense on T2 weighted imaging.
Two patients only had T2 hypointense lesions,
that were consistent with casseous necrosis of the mitral valve.
Signal intensity,
SIR and cardiac tumors
Significant differences in mean and minimum SIb300 (153.17 ± 45.97 vs.
362.02 ± 104.6 and 58.58 ± 18.17 vs.
256.41 ± 89.49; p < 0.05,
respectively) and SIR (1.38±0.58 vs.
2.54±0.7 and 0.88±0.33 vs.
2.87±1.05; p < 0.05,
respectively) between benign and malignant lesions were observed (Table 2; figure 1).
The SImean b300 and SImin b300 of pectoralis muscle was equivalent.
Fig. 3
Fig. 4
Although solid lesions had higher SImean b300,
this difference were not significant between solid and cystic cardiac tumors (242.16 ± 58,52 vs.
102.77 ± 43; p > 0.05,
respectively).
Contrarily significant differences between solid and cystic tumors were obtained in SImin b300 (138.94 ± 41.35 vs.
24.82 ± 9.64; p < 0.05,
respectively),
SIRmean (2.05 ± 0.57 vs.
0.51 ± 0.21; p < 0.05,
respectively) and SIRmin (1.78 ± 0.51 vs.
0.20 ± 0.08; p< 0.05,
respectively),
being higher in solid tumors (table 3; figure 2).
Fig. 5
Fig. 6
Casseous necrosis of the mitral valve had the lowest SImean b300 (65.36 ± 16.26),
SImin b300 (34.28 ± 14.11) and SIRmean (0.42 ± 0.01).
Conversely,
malignant cardiac neoplasms had the highest SImean b300 (362.02 ± 233.91),
SImin b300 (256.41 ± 200.1) and SIRmin (2.87 ± 2.37).
When comparing mean and minimum SI and SIR between etiology subcategories we found statistically significant differences among them (table 4; figure 3).
Fig. 7
Fig. 8
ADC,
ADCr and cardiac tumors
Although having benign lesions higher ADCmean,
ADCmin and ADCrmin higher magnitude compared to malignant ones,
the differences were not significant,
probably due to the small number of patients.
The only significant differences between benign and malignant lesions detected in our cohort was on ADCrmean (3.35 ± 1.15 vs.
0.98 ± 1.17; p < 0.05,
respectively) (table 5; figure 4).
Fig. 9
Fig. 10
Solid lesions had smaller ADC based parameters compared to cystic ones,
although the differences in magnitude were not significant (table 6; figure 5).
Fig. 11
Fig. 12
Casseous necrosis of the mitral valve displayed the lowest ADCmean (1.79 ± 1.31 x 10-3 mm2/s) and ADCmin (0.1 x 10-3 mm2/s) values whereas malignant lesions showed the lowest ADCrmean (0.98 ± 0.39) and ADCrmin (0.87 ± 0.32) (table 4; figure 6).
Fig. 7
All casseous necrosis of the mitral valve showed T2 hypointensity and a T2 “black out” effect in accordance with the lowest ADC values observed.
When different etiologic subcategories were compared,
no significant differences between them were obtained for both SI based and ADC based parameters except for ADCrmin (table 4; figure 6).
Interobserver agreement and ROC analysis
Contrarily to ADC based measures (ICC: 0.345 – 0.681),
SI (ICC: 0.877 – 0.987) and SIR (ICC: 0.672 – 0.941) displayed excellent interobserver reproducibility.
In the ROC analysis SI based parameters showed higher area under the curve (AUC),
compared to ADC based parameters.
Although for SI based measurements SImin b300 showed the highest AUC,
was SIRmean which displayed better diagnostic performance (figure 7; table 7).
Fig. 14
Fig. 15
For that reason,
SIRmean had a 100% sensitivity and 84.6% specificity for differentiating benign from malignant lesions applying a threshold value of 1.32.
Contrarily,
ADCrmean was the ADC based parameter which showed higher AUC on ROC analysis (0.815) and better diagnostic performance for a threshold value of 1.26 (80% and 76.9% for sensitivity and specificity,
respectively) (figure 7; table 7).
Selected cases of cardiac tumors evaluated with DWI:
1.
Casseous necrosis of the mitral valve (CNMV)
Fig. 16
2.
Pericardial cyst
Fig. 17
3.
Myxoma
Fig. 18
4.
Malignant cardiac tumor
Fig. 19
5.
Cardiac thrombus
Fig. 20