Multiparametric Magnetic Resonance Imaging Versus Fine Needle Aspiration Cytology for Parotid Gland Neoplasms

Accuracy of Multiparametric Magnetic Resonance Imaging Versus Fine Needle Aspiration Cytology in the Preoperative Work-Up of Parotid Gland Neoplasms

Parotid gland tumors are mostly treated surgically, but the extent of parotidectomy is decided upon preoperative work-up information. Preoperative management generally includes clinical evaluation, collection of a pathological sample, most often through fine-needle aspiration cytology (FNAC), and imaging. FNAC, despite its high sensitivity and specificity, has the drawback of an approximately 20 per cent rate of nondiagnostic or indeterminate result. Magnetic Resonance Imaging (MRI) provides the best morphological description of the lesion, which is helpful to the surgeon for the planning of the intervention. Recently, advanced functional techniques have been introduced, in association to the conventional morphologic ones: diffusion-weighted imaging (DWI) and dynamic contrast-enhanced imaging (DCEI) demonstrated the ability to provide information about the possible histological origin of parotid lesions. Multiparametric MRI (mp-MRI) comes from the combination of anatomical and functional sequences.

The Authors postulate that mp-MRI evaluation may be able to provide information not only about the extension of the lesion, but also about histology, with a high accuracy, at least comparable to ultrasound-guided FNAC.

In the present study, the Authors aim to define the value of FNAC and mp-MRI in the preoperative management of parotid gland tumors, comparing their success intended as the capability of the exam to be both diagnostic and accurate in formulating the correct histological suspect of malignancy.

Participants are patients affected by parotid gland neoplasms, candidates for surgical resection. The lesion will preoperatively be assessed with both clinical evaluation, ultrasound-guided FNAC and mp-MRI in our Institution. Mp-MRI includes conventional sequences, DWI and DCEI; its interpretation will allow the definition of the suspect histology. FNAC and mp-MRI suspects will be compared to the final histopathological report after surgical removal of the neoplasm.

The study considers a total of 100 patients, of whom 50 are analyzed retrospectively (being already operated after obtaining both FNAC and mp-MRI preoperatively) and the remaining 50 to be enrolled prospectively.

Participants affected by parotid gland neoplasms are evaluated by an ear, nose and throat surgeon in our hospital, who defines the indication to surgical removal.

Preoperative management includes:

• Clinical Evaluation: participant's history and physical examination are collected. Participants are asked for the presence of pain, facial nerve weakness, fixation to skin and surrounding tissues, trismus, skin ulceration, lymphadenopathy, numbness, B symptoms.
• FNAC: it is always performed under US guidance in our Department of Radiology, and subsequently analyzed by a cytopathologist in our Department of Pathology. The same sampling technique is always employed. After skin disinfection, a 23-gauge needle attached to a syringe is inserted into the lesion and moved inside it to obtain enough cytological material. The material is subsequently expelled onto a glass slide, and the smear is prepared between two glass slides and fixed. The procedure is repeated two to three times to reduce the probability of nondiagnostic results.
• mp-MRI: MRI scans are conducted using a 1.5-T imager (Philips Health-care, DA Best, The Netherlands) with a head and neck array coil. The MRI study protocol consists of localizer sequences in transversal, sagittal and coronal planes followed by two axial T2 weighted images: a T2 Turbo Spin-Echo (repetition time -TR-: 4465, echo time -TE-: 80, section thickness: 3mm, intersection gap: 0.3mm) centered on parotid glands and a T2 MV sequence for lymph nodes identification (TR: 3000, TE: 83, section thickness: 4mm, intersection gap: 0.4mm). Then axial T1 weighted images are obtained (TR: 571, TE: 16, section thickness: 3 mm, intersection gap: 0.3mm). Coronal T1, T2 and STIR images are subsequently obtained. Thereafter, DWI is performed by using a multisection spin-echo singleshot echoplanar sequence in the transverse plane (TR: 2800, TE: 160, 20 - 25cm field of view, 92 x 90 matrix, 3mm section thickness). Sensitizing diffusion gradients are applied sequentially in the x, y, and z directions with b values of 0, 500, and 1000 sec/mm2. ADC maps are generated. After detecting the mass in the morphological images, The Radiologist identifies corresponding slice position for a T1 weighted FLASH sequence (TR: 30, TE: 4.6) in the transversal plane, used for DCEI. Each patient receives intravenous contrast injection (weight-adapted dosage, 0.1ml Gadolinium per Kilogram body weight) with a flow rate of 2ml/second using a power injector, followed by a 20ml saline flush. The contrast agent application and the T1-weighted FLASH sequence are started simultaneously. MR images are sequentially obtained at 5sec intervals for 360sec. After DCEI, postcontrast T1 weighted spin echo images are obtained, with the parameters similar to those of the pre-contrast imaging. A region of interest (ROI) for signal intensity measurement is drawn manually avoiding the vessels and necrotic or cystic portions of the tumors. Thereafter, the Radiologist plots the average signal intensity within the ROI against time, and a Time Intensity Curve (TIC) is constructed, using a dedicated software (Philips Intellispace Portal). The ROI can be identified with an adequate accuracy only for tumors with a minimum diameter of 1cm; therefore for smaller lesions a mp-MRI analysis is not possible. TICs are constructed as follows: Time to peak (Tpeak) is manually positioned by the Radiologist in the first value of the curve in which the signal intensity (SI) reached the 90% of the difference between the maximum (SImax) and the starting (SIpre) signals: Tpeak > 0.9 (SImax - SIpre) + SIpre. Washout Ratio (WR) is calculated at 300sec (SIend) and expressed as a percentage, using the following equation: WR = [(SImax - SIend) / (SImax - SIpre)] x 100 (%). TICs are defined as follows: Type A: Tpeak > 120sec (gradual enhancement, ascending plateau); Type B: Tpeak ≤ 120sec and WR ≥ 30% (early enhancement, early washout); Type C: Tpeak ≤ 120sec and WR < 30% (early enhancement, late washout); Type D: flat curve (vacillation of the curve around the low signal level may be observed due to imaging artifacts).

Interpretation of preoperative management findings aims to label each case as "benign" or "malignant", and is performed as follows:

• Clinical Evaluation: malignancy is suspected if any of pain, facial nerve weakness, fixation to skin and surrounding tissues, trismus, skin ulceration, lymphadenopathy, numbness, B symptoms is present.
• FNAC results were classified as: "benign" if a specific histologically benign diagnosis was made, if the report stated "no evidence of malignancy", or if only benign diagnoses were listed in the differential; "malignant" if a specific histologically malignant diagnosis was made, if malignant cells were identified, if the report stated "suspicious for malignancy" or "suspicious for lymphoproliferative disorder". The decision to consider as malignant also reports stating "suspicious for" comes from the observation that surgeons would manage these lesions as malignant anyway; "indeterminate" if, despite the presence of a cytologically adequate sample, the pathologist was not able to define the benign or malignant nature of the lesion; "nondiagnostic" if the pathological sample was insufficient or did not contain enough meaningful material.
• Conventional MRI results are classified as "malignant" if: presence of infiltration of surrounding non-glandular structures (including the facial nerve) is identified; presence of suspect metastatic lymph nodes is identified; the tumor is characterized by low signal on T2-weighted images and displays ill-defined borders. All other cases were classified as "benign".
• Multiparametric MRI results are recorded, and interpreted according to a stepwise approach. The first step of mp-MRI interpretation is the evaluation of conventional MRI findings. If the tumor presented features undoubtedly indicating a malignant origin (infiltration of contiguous non-glandular structures - including the facial nerve - or evidence of lymph node metastases), the lesion is immediately labeled as "malignant" at mp-MRI. The third category used to define malignant nature at conventional MRI evaluation, such as low signal on T2-weighted images and ill-defined borders is not considered as undoubtedly indicating malignancy, and further steps of mp-MRI evaluation are performed. After conventional MRI evaluation, advanced MRI techniques are used. The second step includes DWI with ADC maps generation and measurement. Depending on the ADC value, three main categories are identified: 1. tumors with extremely low ADC value (< 0.6 x 10^-3 mm2/sec); 2. tumors with low or intermediate ADC value (≥ 0.6 x 10^-3 mm2/sec and ≤ 1.4 x 10^-3 mm2/sec);
• tumors with high ADC value (> 1.4 x 10^-3 mm2/sec). The third step is the construction of a TIC using DCEI. Findings obtained with both DWI and DCEI are combined and interpreted as follows: A. high ADC value (> 1.4 x 10^-3 mm2/sec) and TIC type A, B (not plausible) or C are most often suggestive of a Pleomorphic Adenoma (PA), but also of myoepithelioma. Due to the rarity of myoepithelioma and to the fact that its surgical treatment is essentially the same of PA, this eventuality is not considered. TIC type B is not considered a plausible possibility, as this finding is very rarely reported in PAs. In such cases, a hemorrhagic complication after a recently performed FNAC in a Warthin Tumor (WT) should be considered. Tumors in this category are labeled as "benign", with suspected histology of PA; B. high ADC value (> 1.4 x 10^-3 mm2/sec) and TIC type D can be referred to cystic lesions. Cysts are characterized by very high ADC values, being composed of fluid or mucinous content. Tumors in this category are labeled as "benign", with suspected histology of cystic lesion; C. low or intermediate ADC value (≥ 0.6 x 10^-3 mm2/sec and ≤ 1.4 x 10^-3 mm2/sec) and TIC type A or C are typical of malignant epithelial tumors, but also of some atypical adenomas (cellular adenomas, basal cell adenomas). In an attempt to reduce the number of false negative results, tumors in this category are labeled as "malignant". No specific tumor histology can be suspected, due to the large overlap of findings and the high number of parotid gland malignant histotypes; D. low or intermediate ADC value (≥ 0.6 x 10^-3 mm2/sec and ≤ 1.4 x 10^-3 mm2/sec) and TIC type B are typical of WTs. Tumors in this category are labeled as "benign", with suspected histology of WT; low or intermediate ADC value (≥ 0.6 x 10^-3 mm2/sec and ≤ 1.4 x 10-3^mm2/sec) and TIC type D are not plausible, as TIC type D is typical of cysts, which are characterized by very high ADC values. In such cases, one should consider the possibility of repositioning the ROI, as this finding may be related to positioning of the ROI in a necrotic or cystic portion (therefore not informative) of a parotid tumors; E. extremely low ADC value (< 0.6 x 10^-3 mm2/sec) and TIC type A (not plausible), B or C are suggestive of parotid gland lymphomas. Tumors in this category are labeled as "malignant", with suspected histology of lymphoma; F. extremely low ADC value (< 0.6 x 10^-3 mm2/sec) and TIC type D are suggestive of lipomas. Lipomas display typical very low ADC values. Tumors in this category are labeled as "benign", with suspected histology of lipoma; G. mp-MRI IS considered "indeterminate" if the exam is unable to reliably measure ADC or construct a TIC, due to a very recently performed FNAC (less than two weeks) with reported complications (clinical evidence of infection or conventional MRI evidence of intralesional hemorrhage), or intense movement artifacts.

Study power analysis:

The number of participants to be treated is decided taking into account the paired design of the study, to detect any difference in the percentage of success (intended as the capability of the diagnostic test to be diagnostic and to correctly identify malignant tumors) between the two tests (FNAC and mp-MRI), assuming a percentage of participants for whom mp-MRI has success while FNAC does not of 20%, a percentage of participants for whom FNAC has success while mp-MRI does not of 5%, a power of 80% and an alfa error of 5%. These percentage strictly depend on the expected rate of non success due to indeterminate or nondiagnostic results of the tests. Therefore, we defined to be 100 participants the number to be treated.

Statistical analysis:

Number of indeterminate and nondiagnostic results, sensitivity, specificity, positive predictive value, negative predictive value and accuracy of Clinical Evaluation, FNAC, conventional MRI and mp-MRI will be reported. The comparison between success of respectively FNAC and mp-MRI will be investigated with the McNemar test. SPSS software (IBM Corporate, US) will be used. P will be considered to be significant if < 0.05.

Inclusion Criteria:

• Parotid gland tumor, candidate for surgical removal

Exclusion Criteria:

• Non-neoplastic lesions (inflammatory, infectious)
• Lesions smaller than 1 cm (multiparametric MRI analysis not feasible)
• Patients who refuse surgical procedure
• Patients who refuse to take part to the present study