Patient population:
117 patients with parotid gland tumors who had been admitted to the ENT Department in years 2013 - 2014 and subsequently,
after exclusion of contra-indications,
underwent MRI of parotid glands prior to surgical treatment were involved in the study.
With exclusion of examinations with heavy artifacts and patients with no prior biopsy,
100 subjects were selected for further analysis - 60 females (21-88 yo,
mean age 54,5) and 40 males (18-84 yo,
mean age 58,5).
All tumors were located in the parotid glands,
the majority within superficial lobe (84%).
Size of lesions varied from 5 to 48 mm.
15 patients presented with multiple lesions of the same histology - in these cases only the largest tumor was assessed.
All patients underwent FNAC prior to surgery.
Final diagnosis was based on the result of the histopathological examination.
All post-operative histopathological examinations have been performed at The Department of Pathomorphology at our institution.
Table 1 presents post-operative histological diagnosis of evaluated parotid tumors.
Patients gave their written consent.
Approval of The Independent Bio-ethic Committee for Scientific Research of Medical University of Gdansk was granted.
Image acquisition and analysis:
All MRI examinations were performed using a 1.5 T scanner (Magnetom Aera,
Siemens,
Erlangen,
Germany) with the use of a head coil.
Table 2 displays the applied MRI examination protocol.
To obtain contrast-enhanced sequences gadolinium contrast agent - gadobutrol (Gadovist,
Bayer Schering Pharma,
Berlin,
Germany) at standard dose of 0,1 mmol/body weight (0,1 ml/body weight) at a rate of 2-3 ml/sec.
followed by a 20mL saline flush had been administered.
No adverse reactions had been reported.
Diffusion restriction has been first evaluated qualitatively basing on the increase of signal intensity with the growing b factor in DWI sequence,
accompanied by low apparent diffusion coefficient value in ACD map.
ADC values have been then measured on ADC map,
generated in dedicated Siemens software,
by manually placing of a region of interest (ROI) over the tumor area.
DCE-MRI analysis was performed using time-signal intensity curves (TIC) obtained with dedicated Siemens software by placing ROI over the lesion.
In case of heterogeneous lesions,
ROI was carefully placed in order to avoid cystic/necrotic/calcified areas and blood vessels in DCE-MRI analysis.
ROI size varied on average from 3-4 mm.
According to the work by Yabuuchi et al.
with reference to time to peak enhancement (Tpeak) and the wash-out rate (WR),
time-signal intensity curves have been divided into four groups (Fig.
1):
A. Gradual enhancement - Tpeak >120 sec,
WR < 10%,
characteristic of 75% pleomorphic adenomas and other adenomas,
B. Early enhancement - Tpeak <120 sec,
WR > 30%,
characteristic of Warthin’s tumor,
C. Early enhancement - Tpeak <120 sec,
WR < 30%,
characteristic of malignant tumors,
D. No enhancement - flat curve,
characteristic of cystic lesions.
Obtained data has been evaluated by two independent radiologists with prior experience in Head and Neck Radiology blinded to clinical history and results of both FNAC and histopathological reports.
Number and location of lesions as well as their morphology,
including the size of the tumor,
signal intensity on T2-,
T1-weighted and T1-weighted images with fat saturation and homogeneity have been assessed.
Subsequently ADC values of the tumor and healthy parotid glands have been measured and enhancement pattern has been assessed with calculation of TICs in order to establish radiological diagnosis.
Presence of lymphadenopathy and signs of perineural spread have been reported as well.
Statistical analysis:
Diagnoses established in FNAC,
radiological and histopathological assessment have been grouped into 11 homogeneous groups for purposes of statistical analysis. Statistical analysis has been performed with use of commercially available software Stata 12. Diagnostic accuracy of FNAC and MRI with inclusion of DWI and post-contrast dynamic imaging were compared with histopathological results obtained from surgical specimens,
which served as a gold reference.
Interobserver agreement in tumor characterization as well as accuracy of diagnosis of each of the observers’ with histopathologic examination have been evaluated. Data was analyzed using the kappa statistics.
Kappa values were calculated according to Cohen.
Kappa’s values range from -1 to +1.
-1 stands for maximal disagreement,
0 means that observed agreement equals chance agreement,
while +1 corresponds to maximal agreement beyond chance.
Kappa (κ) values may also be interpreted as a percentage of agreement between observers.
Level of agreement was measured according to Fleiss [9]: κ <0.40 (fair agreement),
κ 0,40 - 0.74 (moderate and good agreement),
κ > 0.75 (excellent agreement) [10].