- Email: [email protected]
The significance of CT scan or MRI in the evaluation of salivary gland tumors
Auris Nasus Larynx 25 (1998) 397 – 402
The significance of CT scan or MRI in the evaluation of salivary gland tumors Kwang Hyun Kim a,*, Myung-Whun Sung a, Ja Bock Yun a, Moon Hee Han b, Chung-Hwan Baek c, Kwang-Chol Chu c, Jae Ho Kim d, Kwang-Sun Lee d a
Department of Otolaryngology —Head and Neck Surgery, College of Medicine, Seoul National Uni6ersity, 28 Yongon-Dong, Chongno-Gu, Seoul 110 -744, South Korea b Department of Diagnostic Radiology, College of Medicine, Seoul National Uni6ersity, Seoul 110 -744, South Korea c Department of Otolaryngology, Samsung Medical Center, College of Medicine, Sungkyunkwan Uni6ersity, Seoul, South Korea d Department of Otolaryngology, College of Medicine, Ulsan Uni6ersity, Seoul, South Korea Received 4 August 1997; received in revised form 15 October 1997; accepted 14 November 1997
Abstract Imaging modalities such as CT scan or MRI are frequently employed for the diagnosis of neoplastic lesions in the salivary glands. To evaluate the efficacy of the CT scan and the MRI in differentiating malignant neoplasm from benign lesions, 120 CT scans and 31 MRIs were retrospectively analyzed from 147 patients with salivary gland masses. All images were analyzed focusing on the presence of several relevant features. The pathologic results were matched with radiological features and also tabulated with radiological assessment. For the CT scans, the contour and margin of the lesion and tissue plane obliteration were found to be statistically significant indicators for malignant neoplasms. Among 69 CT scans interpreted as ‘benign’ by a radiologist, five cases (7%) were histologically diagnosed as ‘malignant’. On the other hand, 20 out of 51 CT scans (39%) were misinterpreted as ‘malignant’. For MRI, two out of 14 cases (14%) were radiologically misdiagnosed as ‘benign’ and six out of 17 patients (35%) as ‘malignant’. In conclusion, whereas both the CT and MRI showed a similar level of accuracy in evaluation of salivary gland tumors, they showed a considerable tendency of misdiagnosis, especially by interpreting benign tumors as ‘malignant’. © 1998 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Salivary gland cancer; Computed tomography; Magnetic resonance imaging
1. Introduction * Corresponding author. Tel.: + 82 2 7602448; fax: + 82 2 7452387.
Imaging modalities such as CT scan or MRI are frequently employed in the evaluation of tu-
0385-8146/98/$ - see front matter © 1998 Elsevier Science Ireland Ltd. All rights reserved. PII S0385-8146(98)00012-1
398
K.H. Kim et al. / Auris Nasus Larynx 25 (1998) 397–402
mors occurring in the salivary glands. However, the efficacy of these imaging modalities in assessing the nature of the mass are still not well-defined. For simple mass lesions in the parotid or submandibular glands, a physical examination might be enough for the decision of operation and the final diagnosis would be made by histologic examination of the surgical specimens. The CT scan and/or MRI might be more useful to evaluate salivary gland masses with signs of suggesting malignancy, such as fixation to the adjacent tissues or facial paralysis, for better surgical planning. Imaging techniques can provide important information about the extent and location of the mass [1]. However, even in cases of clinically suspected malignancy, radiologists may not be able to comment on if the lesion is malignant or not. Furthermore, some cases of discrepancy have been seen between the radiological opinions and histologic diagnoses, which triggered several questions; to what extent can CT scan or MRI play a role in differentiating malignant tumors from benign lesions? What are characteristic findings suggesting malignancy in these modalities? Which modality should be chosen to evaluate the masses in the salivary glands for preoperative evaluation? This study has tried to answer these questions by a direct comparison of radiological interpretations and histological diagnoses.
contrast infusions. The MRI examinations were also performed using a 1.5 T MR scanner (Siemens Magnetom, Siemens AG). The Axial T2-weighted images, axial and coronal T1-weighted images with 5 mm thickness were acquired at 12 mm intervals. The gadolinium-enhanced T1-weighted axial image was also obtained in axial and coronal planes. The images were analyzed with special concerns on the following features for CT scans: (a) the contour and margin of the mass; (b) the pattern of enhancement; (c) the presence of cystic change; (d) necrosis/hemorrhage; (e) calcification; or (f) tissue plane obliteration. The signal intensity was also evaluated for the RI. The contour of the mass was defined as smooth, lobulated or irregular and the margin as clear or poor. Among the low attenuated resonance images in MRI, the images were regarded with a thin and smooth wall as cyst and with a indefinite margin or irregular wall as necrosis. The signal intensity of the tumor was compared with that of normal salivary gland tissue. CT scans and MRI were interpreted by one radiologist specialized in head and neck imaging, without any information on the pathologic reports. The pathologic results were compared with each of the radiologic imaging features and tabulated with the radiological assessments. The results were statistically analyzed by the x 2-test.
3. Results 2. Materials and methods
3.1. Histologic findings The CT scans or MRIs of 147 cases were retrospectively analyzed whose diagnoses were histologically confirmed. There were 120 patients with CT scans, 31 patients with MRI and four patients with both of them. The locations of the tumors were as follows: (a) 116 parotid glands; (b) 23 submandibular glands; and (c) eight minor salivary glands. The CT scans were performed using various scanners (GE HiSpeed, GE 9800Q, GE Medical Systems, Milwaukee, WI and Siemens Somatom Plus-S, Siemens AG, Elangen, Germany). The axial and coronal images with a 3 mm thickness were obtained at 3 mm intervals in the parotid area and at 5 mm intervals in the remainder of the neck after
On reviewing 147 patients, 136 were found to have neoplastic lesions (91 benign and 45 malignant tumors). The others were seven inflammatory disease and four non-neoplastic cysts. Histologic diagnoses in the group of benign tumors were as follows: (a) pleomorphic adenoma (n=63, including five recurrent cases); (b) Warthin’s tumor (n=23); (c) monomorphic adenoma (n= 3) and others. Mucoepidermoid carcinoma was the most frequent primary malignant salivary gland tumor (n= 11) and acinic cell carcinoma (n= 7) and adenoid cystic carcinoma (n= 7) were the next. In two of the cases metastatic adenocarcinoma was revealed (Table 1).
K.H. Kim et al. / Auris Nasus Larynx 25 (1998) 397–402
A total of 27% of parotid neoplasms were found to be histologically malignant and 47% of submandibular masses were malignant.
3.2. Imaging features All images were checked for the presence of relevant imaging features, which were tabulated with histologic assessment of ‘benign’ or ‘malignant’ (Tables 2 and 3). For CT scans, among eight features evaluated, the contour of tumor, the margin and tissue plane obliteration were found to be statistically significant indicators of malignant neoplasm (P B0.01). Other findings, such as the pattern of contrast enhancement, cystic degeneration, calcification and necrosis/ hemorrhage did not show any statistical significance in differentiating ‘benign’ from ‘malignant’ lesions. On reviewing MRI, the irregular contour of the lesion significantly favored malignancy (P B Table 1 Distribution of a histologic diagnosis in 147 patients with salivary gland tumors Tumours
Pathologic diagnosis
Benign (n =91)
Pleomorphic adenoma 58 Warthin’s tumor 23 Recurred pleomorphic 5 adenoma Monomorphic adenoma 3 Myoepithelial tumor 1 Neurilemmoma 1 Mucoepidermoid 11 carcinoma Acinic cell carcinoma 7 Adenoid cystic carcinoma 7 Carcinoma ex pleomorphic 4 adenoma Malignant mixed tumor 2 Metastatic adenocarcinoma 2 Epithelial-myoepithelial 1 carcinoma Others 7 Chronic inflammation 7
Malignant (n = 45)
Inflammations (n= 7) Non-neoplastic lesions (n =4)
Number
Lymphoepithelial cyst
2
Lymphangioma
2
399
0.01) and intermediate to low signal intensity in T2-weighted images seemed to be indicative of malignancy (P= 0.052). Based on these findings, this study tried to determine the efficacy of the CT scan or MRI in the differential diagnosis of salivary gland tumor. The histologic diagnoses were matched with the interpretations of the radiologist. If any one of the imaging features which favored malignancy with a statistical significance (Table 2), i.e. a lobular or irregular contour, a poor margin of the tumor or tissue plane obliteration was found, a radiologic assessment was made as ‘malignant’. By applying this criteria to all patients with CT scans, 69 patients were interpreted as ‘benign’ in radiological assessment but five (7%) out of them turned out to be ‘malignant’ by histologic examination (Table 4). These patients included each of acinic cell carcinoma, adenoid cystic carcinoma, epithelial-myoepithelial carcinoma, basal cell carcinoma and lymphoma. On the other hand, 20 patients (39%) had histologically benign lesions, in spite of ‘malignant’ interpretation of CT scans. The majority of these cases were pleomorphic adenoma (12 cases), including one recurrent lesion. The others were Warthin’s tumor (two cases), chronic inflammation (five cases) and one lymphangioma. Similarly, with MRI misdiagnosed two out of 14 cases (14%) as ‘benign’ and six out of 17 patients (35%) as ‘malignant’, by applying the same criteria (Table 5). The former two cases were acinic cell carcinoma and mucoepidermoid carcinoma. The latter six cases were all pleomorphic adenomas, including two recurrent cases. In the study, five recurrent pleomorphic adenomas were enrolled and the CT scan or MRI misdiagnosed three out of them as malignant. The sensitivity and specificity of CT scans for detecting malignancy were 0.93 and 0.61, respectively. MRI showed 0.83 for sensitivity and 0.63 for specificity. MRI seemed to provide better images in distinguishing the tumors from the surrounding tissue when normal parenchyma of the salivary gland showed high back ground intensity.
400
K.H. Kim et al. / Auris Nasus Larynx 25 (1998) 397–402
Table 2 Imaging features of CT scans in benign and malignant salivary gland neoplasms (n = 109)a Features
Variables
Benign (n =73)
Malignant (n =36)
P-value*
Contour
Smooth Lobulated Irregular Clear Poor Present Absent Rim Solid Present Absent Present Absent Present Absent
53 18 2 63 10 6 67 15 58 16 57 25 48 5 68
13 12 11 19 17 22 14 4 32 8 28 17 19 4 32
B0.01
Margin Tissue plane obliteration Enhancement pattern Cystic change Necrosis or hemorrhage Calcification
B0.01 B0.01 0.34 0.97 0.27 0.70
a
Inflammations and non-neoplastic lesions were excluded. * x 2-test.
4. Discussion In this study, both of CT scan and MRI were excellent in delineating the lesions. Both modalities can show us the exact intra- or extraglanduar location of the tumor or extension into the adjacent parapharyngeal space. Some studies have suggested MRI was the choice for evaluation of major salivary gland tumor [2 – 4] because MRI can provide images of coronal and sagittal orientations, which is one of the advantages of this modality over CT scan. According to the results of this study, however, CT scan was not at all inferior to MRI in assessing the nature of the tumor in the salivary glands. It was found that radiologic features indicating malignancy with a statistical significance were similar in both of CT scan and MRI, i.e. a lobular or irregular contour of the mass, a poor tumor margin or obliteration of the adjacent tissue plane, as observed in earlier reports [5]. An additional feature from MRI that can be used for differential diagnosis is signal intensity on T2-weighted images. Joe and Westesson indicated that carcinomas showed intermediate to low signal intensity, whereas
pleomorphic adenomas had high signal intensity [6]. Som et al. [7] also pointed out that the characteristic findings of high grade malignancy of the parotid were poorly defined margins and low signal intensities in T1- and T2-weighted images. This study also revealed a higher percentage of intermediate to low signal intensity in T2-weighted images in malignant tumors (six out of 12) but these figures were not statistically significant. The data demonstrated that imaging features cannot be relied on as a sole preoperative determinant for differentiation of malignancy from benign tumors. When the radiological interpretation was made as ‘malignant’ for cases with any one of the imaging features favoring malignancy from Table 2, both imaging techniques showed similarly high tendencies of misdiagnosis, especially by misinterpreting ‘benign’ tumor as ‘malignant’ one. The misdiagnosis rates were upto 39% for CT scans and 35% for MRI. However, even though imaging alone cannot lead to histologic differential diagnosis, these are one of the essential work-ups for salivary glands tumors, especially when the location or extent of the mass is doubtful.
K.H. Kim et al. / Auris Nasus Larynx 25 (1998) 397–402
401
Table 3 Imaging features of MRI in benign and malignant salivary gland neoplasms (n =30)a Featues Contour
Smooth Lobulated Irregular Present Absent Present Absent Present Absent Present Absent T1BT2 T1= T2 T1\T2 Rim Solid High Intermediate High Intermediate High/Low Low
Cystic change Necrosis Tissue plane obliteration Signal void Margin clearance
Enhancement pattern T1-weighted image T2-weighted
Benign (n = 18)
Malignant (n =12)
P-value*
12 6 0 3 15 7 9 3 15 0 18 6 3 9 1 17 5 13 16 0 0 2
7 1 4 1 11 1 11 5 7 1 11 4 2 6 0 12 2 10 6 2 2 2
B0.01
0.98 0.10 0.21 0.84 1.00
0.40 0.48 0.05
a
Inflammations and non-neoplastic lesions were excluded. * x 2-test.
Since this data demonstrated both imaging techniques had similar levels of accuracy in evaluation of salivary gland tumors, especially in the differential diagnosis of malignancy from benign masses, it was suggested that the CT scan might be considered as an initial choice of imaging
Table 4 Discrepancy between CT readings and pathologic diagnosis (n=110) Pathologist’s diagnosis
Malignant Benign Neoplasms Inflammations Non-neoplastic lesions
Radiologist’s diagnosis Malignant (n= 51)
Benign (n = 69)
31 20 14 5 1
5 64 59 2 3
modality in respect of cost-effectiveness for workup of mass lesions in the salivary glands. As an extension of this study, a scoring system might be developed, by giving a proper weight to each item of findings, which might be able to reduce the high misdiagnosis tendency of interpreting benign tumors as ‘malignant’.
Table 5 Discrepancy between MRI readings and pathologic diagnoses (n =31) Pathologist’s diagnosis
Malignant Benign Neoplasms Non-neoplastic lesions
Radiologist’s diagnosis Malignant (n =17)
Benign (n = 14)
10 7 6 1
2 12 12 0
402
K.H. Kim et al. / Auris Nasus Larynx 25 (1998) 397–402 Hanafee WN. Parotid gland: plain and gadolinium-enhanced MR imaging. Radiology 1987;163:405 – 9. [5] Freling MJM, Molenaar WM, Vermey A, Mooyaart EL, Panders AK, Annyas AA, Thijn CJ. Malignant parotid tumors: clinial use of MR imaging and histologic correlation. Radiology 1992;185:691 – 6. [6] Joe VQ, Westesson P. Tumor of the parotid gland: MRI characteristics of various histologic types. Am J Roentgenol 1994;163:433 – 8. [7] Som PM, Biller HF. High-grade malignancies of the parotid gland: identification with MR imaging. Radiology 1989;173:823 – 6.
References [1] Barsotti JB, Westesson P, Coniglio JU. Superiority of MR over CT for imaging parotid tumor. Ann Otol Rhinol Laryngol 1994;103:737–40. [2] Casselman JW, Mancusso AA. Major salivary gland masses: comparison of MR imaging and CT. Radiology 1987;165:183 – 9. [3] Mandelblatt SM, Braun IF, Davis PC, Fry SM, Jacbs LH, Hoffman JC. Parotid masses: MR imaging. Radiology 1987;163:411 – 4. [4] Teresei LM, Lufkin RB, Wortham DG, Abemayor E,
.
The significance of CT scan or MRI in the evaluation of salivary gland tumors Kwang Hyun Kim a,*, Myung-Whun Sung a, Ja Bock Yun a, Moon Hee Han b, Chung-Hwan Baek c, Kwang-Chol Chu c, Jae Ho Kim d, Kwang-Sun Lee d a
Department of Otolaryngology —Head and Neck Surgery, College of Medicine, Seoul National Uni6ersity, 28 Yongon-Dong, Chongno-Gu, Seoul 110 -744, South Korea b Department of Diagnostic Radiology, College of Medicine, Seoul National Uni6ersity, Seoul 110 -744, South Korea c Department of Otolaryngology, Samsung Medical Center, College of Medicine, Sungkyunkwan Uni6ersity, Seoul, South Korea d Department of Otolaryngology, College of Medicine, Ulsan Uni6ersity, Seoul, South Korea Received 4 August 1997; received in revised form 15 October 1997; accepted 14 November 1997
Abstract Imaging modalities such as CT scan or MRI are frequently employed for the diagnosis of neoplastic lesions in the salivary glands. To evaluate the efficacy of the CT scan and the MRI in differentiating malignant neoplasm from benign lesions, 120 CT scans and 31 MRIs were retrospectively analyzed from 147 patients with salivary gland masses. All images were analyzed focusing on the presence of several relevant features. The pathologic results were matched with radiological features and also tabulated with radiological assessment. For the CT scans, the contour and margin of the lesion and tissue plane obliteration were found to be statistically significant indicators for malignant neoplasms. Among 69 CT scans interpreted as ‘benign’ by a radiologist, five cases (7%) were histologically diagnosed as ‘malignant’. On the other hand, 20 out of 51 CT scans (39%) were misinterpreted as ‘malignant’. For MRI, two out of 14 cases (14%) were radiologically misdiagnosed as ‘benign’ and six out of 17 patients (35%) as ‘malignant’. In conclusion, whereas both the CT and MRI showed a similar level of accuracy in evaluation of salivary gland tumors, they showed a considerable tendency of misdiagnosis, especially by interpreting benign tumors as ‘malignant’. © 1998 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Salivary gland cancer; Computed tomography; Magnetic resonance imaging
1. Introduction * Corresponding author. Tel.: + 82 2 7602448; fax: + 82 2 7452387.
Imaging modalities such as CT scan or MRI are frequently employed in the evaluation of tu-
0385-8146/98/$ - see front matter © 1998 Elsevier Science Ireland Ltd. All rights reserved. PII S0385-8146(98)00012-1
398
K.H. Kim et al. / Auris Nasus Larynx 25 (1998) 397–402
mors occurring in the salivary glands. However, the efficacy of these imaging modalities in assessing the nature of the mass are still not well-defined. For simple mass lesions in the parotid or submandibular glands, a physical examination might be enough for the decision of operation and the final diagnosis would be made by histologic examination of the surgical specimens. The CT scan and/or MRI might be more useful to evaluate salivary gland masses with signs of suggesting malignancy, such as fixation to the adjacent tissues or facial paralysis, for better surgical planning. Imaging techniques can provide important information about the extent and location of the mass [1]. However, even in cases of clinically suspected malignancy, radiologists may not be able to comment on if the lesion is malignant or not. Furthermore, some cases of discrepancy have been seen between the radiological opinions and histologic diagnoses, which triggered several questions; to what extent can CT scan or MRI play a role in differentiating malignant tumors from benign lesions? What are characteristic findings suggesting malignancy in these modalities? Which modality should be chosen to evaluate the masses in the salivary glands for preoperative evaluation? This study has tried to answer these questions by a direct comparison of radiological interpretations and histological diagnoses.
contrast infusions. The MRI examinations were also performed using a 1.5 T MR scanner (Siemens Magnetom, Siemens AG). The Axial T2-weighted images, axial and coronal T1-weighted images with 5 mm thickness were acquired at 12 mm intervals. The gadolinium-enhanced T1-weighted axial image was also obtained in axial and coronal planes. The images were analyzed with special concerns on the following features for CT scans: (a) the contour and margin of the mass; (b) the pattern of enhancement; (c) the presence of cystic change; (d) necrosis/hemorrhage; (e) calcification; or (f) tissue plane obliteration. The signal intensity was also evaluated for the RI. The contour of the mass was defined as smooth, lobulated or irregular and the margin as clear or poor. Among the low attenuated resonance images in MRI, the images were regarded with a thin and smooth wall as cyst and with a indefinite margin or irregular wall as necrosis. The signal intensity of the tumor was compared with that of normal salivary gland tissue. CT scans and MRI were interpreted by one radiologist specialized in head and neck imaging, without any information on the pathologic reports. The pathologic results were compared with each of the radiologic imaging features and tabulated with the radiological assessments. The results were statistically analyzed by the x 2-test.
3. Results 2. Materials and methods
3.1. Histologic findings The CT scans or MRIs of 147 cases were retrospectively analyzed whose diagnoses were histologically confirmed. There were 120 patients with CT scans, 31 patients with MRI and four patients with both of them. The locations of the tumors were as follows: (a) 116 parotid glands; (b) 23 submandibular glands; and (c) eight minor salivary glands. The CT scans were performed using various scanners (GE HiSpeed, GE 9800Q, GE Medical Systems, Milwaukee, WI and Siemens Somatom Plus-S, Siemens AG, Elangen, Germany). The axial and coronal images with a 3 mm thickness were obtained at 3 mm intervals in the parotid area and at 5 mm intervals in the remainder of the neck after
On reviewing 147 patients, 136 were found to have neoplastic lesions (91 benign and 45 malignant tumors). The others were seven inflammatory disease and four non-neoplastic cysts. Histologic diagnoses in the group of benign tumors were as follows: (a) pleomorphic adenoma (n=63, including five recurrent cases); (b) Warthin’s tumor (n=23); (c) monomorphic adenoma (n= 3) and others. Mucoepidermoid carcinoma was the most frequent primary malignant salivary gland tumor (n= 11) and acinic cell carcinoma (n= 7) and adenoid cystic carcinoma (n= 7) were the next. In two of the cases metastatic adenocarcinoma was revealed (Table 1).
K.H. Kim et al. / Auris Nasus Larynx 25 (1998) 397–402
A total of 27% of parotid neoplasms were found to be histologically malignant and 47% of submandibular masses were malignant.
3.2. Imaging features All images were checked for the presence of relevant imaging features, which were tabulated with histologic assessment of ‘benign’ or ‘malignant’ (Tables 2 and 3). For CT scans, among eight features evaluated, the contour of tumor, the margin and tissue plane obliteration were found to be statistically significant indicators of malignant neoplasm (P B0.01). Other findings, such as the pattern of contrast enhancement, cystic degeneration, calcification and necrosis/ hemorrhage did not show any statistical significance in differentiating ‘benign’ from ‘malignant’ lesions. On reviewing MRI, the irregular contour of the lesion significantly favored malignancy (P B Table 1 Distribution of a histologic diagnosis in 147 patients with salivary gland tumors Tumours
Pathologic diagnosis
Benign (n =91)
Pleomorphic adenoma 58 Warthin’s tumor 23 Recurred pleomorphic 5 adenoma Monomorphic adenoma 3 Myoepithelial tumor 1 Neurilemmoma 1 Mucoepidermoid 11 carcinoma Acinic cell carcinoma 7 Adenoid cystic carcinoma 7 Carcinoma ex pleomorphic 4 adenoma Malignant mixed tumor 2 Metastatic adenocarcinoma 2 Epithelial-myoepithelial 1 carcinoma Others 7 Chronic inflammation 7
Malignant (n = 45)
Inflammations (n= 7) Non-neoplastic lesions (n =4)
Number
Lymphoepithelial cyst
2
Lymphangioma
2
399
0.01) and intermediate to low signal intensity in T2-weighted images seemed to be indicative of malignancy (P= 0.052). Based on these findings, this study tried to determine the efficacy of the CT scan or MRI in the differential diagnosis of salivary gland tumor. The histologic diagnoses were matched with the interpretations of the radiologist. If any one of the imaging features which favored malignancy with a statistical significance (Table 2), i.e. a lobular or irregular contour, a poor margin of the tumor or tissue plane obliteration was found, a radiologic assessment was made as ‘malignant’. By applying this criteria to all patients with CT scans, 69 patients were interpreted as ‘benign’ in radiological assessment but five (7%) out of them turned out to be ‘malignant’ by histologic examination (Table 4). These patients included each of acinic cell carcinoma, adenoid cystic carcinoma, epithelial-myoepithelial carcinoma, basal cell carcinoma and lymphoma. On the other hand, 20 patients (39%) had histologically benign lesions, in spite of ‘malignant’ interpretation of CT scans. The majority of these cases were pleomorphic adenoma (12 cases), including one recurrent lesion. The others were Warthin’s tumor (two cases), chronic inflammation (five cases) and one lymphangioma. Similarly, with MRI misdiagnosed two out of 14 cases (14%) as ‘benign’ and six out of 17 patients (35%) as ‘malignant’, by applying the same criteria (Table 5). The former two cases were acinic cell carcinoma and mucoepidermoid carcinoma. The latter six cases were all pleomorphic adenomas, including two recurrent cases. In the study, five recurrent pleomorphic adenomas were enrolled and the CT scan or MRI misdiagnosed three out of them as malignant. The sensitivity and specificity of CT scans for detecting malignancy were 0.93 and 0.61, respectively. MRI showed 0.83 for sensitivity and 0.63 for specificity. MRI seemed to provide better images in distinguishing the tumors from the surrounding tissue when normal parenchyma of the salivary gland showed high back ground intensity.
400
K.H. Kim et al. / Auris Nasus Larynx 25 (1998) 397–402
Table 2 Imaging features of CT scans in benign and malignant salivary gland neoplasms (n = 109)a Features
Variables
Benign (n =73)
Malignant (n =36)
P-value*
Contour
Smooth Lobulated Irregular Clear Poor Present Absent Rim Solid Present Absent Present Absent Present Absent
53 18 2 63 10 6 67 15 58 16 57 25 48 5 68
13 12 11 19 17 22 14 4 32 8 28 17 19 4 32
B0.01
Margin Tissue plane obliteration Enhancement pattern Cystic change Necrosis or hemorrhage Calcification
B0.01 B0.01 0.34 0.97 0.27 0.70
a
Inflammations and non-neoplastic lesions were excluded. * x 2-test.
4. Discussion In this study, both of CT scan and MRI were excellent in delineating the lesions. Both modalities can show us the exact intra- or extraglanduar location of the tumor or extension into the adjacent parapharyngeal space. Some studies have suggested MRI was the choice for evaluation of major salivary gland tumor [2 – 4] because MRI can provide images of coronal and sagittal orientations, which is one of the advantages of this modality over CT scan. According to the results of this study, however, CT scan was not at all inferior to MRI in assessing the nature of the tumor in the salivary glands. It was found that radiologic features indicating malignancy with a statistical significance were similar in both of CT scan and MRI, i.e. a lobular or irregular contour of the mass, a poor tumor margin or obliteration of the adjacent tissue plane, as observed in earlier reports [5]. An additional feature from MRI that can be used for differential diagnosis is signal intensity on T2-weighted images. Joe and Westesson indicated that carcinomas showed intermediate to low signal intensity, whereas
pleomorphic adenomas had high signal intensity [6]. Som et al. [7] also pointed out that the characteristic findings of high grade malignancy of the parotid were poorly defined margins and low signal intensities in T1- and T2-weighted images. This study also revealed a higher percentage of intermediate to low signal intensity in T2-weighted images in malignant tumors (six out of 12) but these figures were not statistically significant. The data demonstrated that imaging features cannot be relied on as a sole preoperative determinant for differentiation of malignancy from benign tumors. When the radiological interpretation was made as ‘malignant’ for cases with any one of the imaging features favoring malignancy from Table 2, both imaging techniques showed similarly high tendencies of misdiagnosis, especially by misinterpreting ‘benign’ tumor as ‘malignant’ one. The misdiagnosis rates were upto 39% for CT scans and 35% for MRI. However, even though imaging alone cannot lead to histologic differential diagnosis, these are one of the essential work-ups for salivary glands tumors, especially when the location or extent of the mass is doubtful.
K.H. Kim et al. / Auris Nasus Larynx 25 (1998) 397–402
401
Table 3 Imaging features of MRI in benign and malignant salivary gland neoplasms (n =30)a Featues Contour
Smooth Lobulated Irregular Present Absent Present Absent Present Absent Present Absent T1BT2 T1= T2 T1\T2 Rim Solid High Intermediate High Intermediate High/Low Low
Cystic change Necrosis Tissue plane obliteration Signal void Margin clearance
Enhancement pattern T1-weighted image T2-weighted
Benign (n = 18)
Malignant (n =12)
P-value*
12 6 0 3 15 7 9 3 15 0 18 6 3 9 1 17 5 13 16 0 0 2
7 1 4 1 11 1 11 5 7 1 11 4 2 6 0 12 2 10 6 2 2 2
B0.01
0.98 0.10 0.21 0.84 1.00
0.40 0.48 0.05
a
Inflammations and non-neoplastic lesions were excluded. * x 2-test.
Since this data demonstrated both imaging techniques had similar levels of accuracy in evaluation of salivary gland tumors, especially in the differential diagnosis of malignancy from benign masses, it was suggested that the CT scan might be considered as an initial choice of imaging
Table 4 Discrepancy between CT readings and pathologic diagnosis (n=110) Pathologist’s diagnosis
Malignant Benign Neoplasms Inflammations Non-neoplastic lesions
Radiologist’s diagnosis Malignant (n= 51)
Benign (n = 69)
31 20 14 5 1
5 64 59 2 3
modality in respect of cost-effectiveness for workup of mass lesions in the salivary glands. As an extension of this study, a scoring system might be developed, by giving a proper weight to each item of findings, which might be able to reduce the high misdiagnosis tendency of interpreting benign tumors as ‘malignant’.
Table 5 Discrepancy between MRI readings and pathologic diagnoses (n =31) Pathologist’s diagnosis
Malignant Benign Neoplasms Non-neoplastic lesions
Radiologist’s diagnosis Malignant (n =17)
Benign (n = 14)
10 7 6 1
2 12 12 0
402
K.H. Kim et al. / Auris Nasus Larynx 25 (1998) 397–402 Hanafee WN. Parotid gland: plain and gadolinium-enhanced MR imaging. Radiology 1987;163:405 – 9. [5] Freling MJM, Molenaar WM, Vermey A, Mooyaart EL, Panders AK, Annyas AA, Thijn CJ. Malignant parotid tumors: clinial use of MR imaging and histologic correlation. Radiology 1992;185:691 – 6. [6] Joe VQ, Westesson P. Tumor of the parotid gland: MRI characteristics of various histologic types. Am J Roentgenol 1994;163:433 – 8. [7] Som PM, Biller HF. High-grade malignancies of the parotid gland: identification with MR imaging. Radiology 1989;173:823 – 6.
References [1] Barsotti JB, Westesson P, Coniglio JU. Superiority of MR over CT for imaging parotid tumor. Ann Otol Rhinol Laryngol 1994;103:737–40. [2] Casselman JW, Mancusso AA. Major salivary gland masses: comparison of MR imaging and CT. Radiology 1987;165:183 – 9. [3] Mandelblatt SM, Braun IF, Davis PC, Fry SM, Jacbs LH, Hoffman JC. Parotid masses: MR imaging. Radiology 1987;163:411 – 4. [4] Teresei LM, Lufkin RB, Wortham DG, Abemayor E,
.