- Email: [email protected]
Staging Papillary Carcinoma of the Thyroid: Magnetic Resonance Imaging vs Ultrasound of the Neck
Clinical Radiology (2000) 55, 222–226 doi:10.1053/crad.1999.0373, available online at http://www.idealibrary.com on
Staging Papillary Carcinoma of the Thyroid: Magnetic Resonance Imaging vs Ultrasound of the Neck A. D. KING, A. T . AHU JA, E . W . H. T O* , G. M . K. T SE† , C. MET R E W E L I Departments of Diagnostic Radiology & Organ Imaging, *Surgery and †Anatomical and Cellular Pathology, The Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, NT, Hong Kong Received: 18 April 1999 Revised: 27 July 1999
Accepted: 20 August 1999
AIM: To evaluate the role of magnetic resonance imaging (MRI) for staging local disease and lymph node metastases in papillary carcinoma of the thyroid by comparing MRI with ultrasound (US) of the neck. MATERIALS AND METHODS: Fourteen patients with papillary carcinoma underwent MRI and US. The images were prospectively reviewed for (1) identification of the primary lesion; (2) presence of multifocal thyroid disease; (3) extracapsular extension; (4) invasion into the trachea, oesophagus and major vessels; and (5) presence of lymph node metastases. Correlation was made with the surgical findings. RESULTS: The site of the primary lesion was correctly identified by US in 14 of 14 (100%) and by MRI in 13 of 14 (93%) of patients. Multifocal thyroid tumour was correctly identified by US in two of two patients (100%) and by MRI in zero of two (0%). Extracapsular extension (n ¼ 9), oesophageal invasion (n ¼ 1) and tracheal invasion (n ¼ 2) were identified by MRI in seven, zero, and one, and by US in six, zero and zero patients, respectively. Invasion of the major vessels was not seen. Metastatic cervical nodes were present in 19 nodal groups in 10 patients involving the internal jugular chain (n ¼ 10), posterior triangle (n ¼ 4) supraclavicular fossa (n ¼ 1) and central group (n ¼ 4). Both MRI and US failed to identify metastatic nodes in the central group but correctly identified 14 of the 15 nodal groups outside the central group. CONCLUSION: Ultrasound should be used as the first line of investigation for detecting the primary lesion, multifocal disease and cervical lymphadenopathy. In cases where the primary tumour is not surrounded by normal thyroid tissue MR imaging should be added to assess extracapsular spread, especially into the trachea. King, A. D. et al. (2000). Clinical Radiology 55, 222–226. q 2000 The Royal College of Radiologists Key words: MR imaging, ultrasound, thyroid, cancer.
Magnetic resonance imaging (MRI) is being used increasingly to image the thyroid gland [1–7] but as yet its role in diagnosis and management of the patient is unclear, especially when compared to the more established techniques for examining the thyroid, such as ultrasound (US). The aim of the study, therefore, was to compare MRI with US of the neck in the staging of papillary carcinoma of the thyroid. Papillary carcinoma was studied, not only because is it the most common thyroid malignancy, but also because imaging can influence the extent of surgical resection by identifying multifocal disease, extracapsular spread including invasion of the aerodigestive tract and regional lymph-node metastases. METHODS
Fourteen consecutive patients (six men, eight women; age range 16–81 years, mean 54 years) presenting with a thyroid Author for correspondence and guarantor of study: Dr A. D. King, Department of Diagnostic Radiology & Organ Imaging, Prince of Wales Hospital, Shatin, NT, Hong Kong. 0009-9260/00/030222+05 $35.00/0
mass confirmed to be a papillary carcinoma on fine needle aspiration (FNA) were entered into the study. Real-time ultrasound images were obtained using a 5–12 MHZ transducer (HDI 5000; Advanced Technology Laboratories, Bothell, WA). MR images were obtained on a 1.5 MRI unit (Philips Gyroscan, Best, Netherlands) using a neck coil (23 cm) and a surface coil. All patients underwent an axial dual turbo spin-echo (repetition time of 2500 ms, echo time of 35 ms, echo train length of 8, 23 cm field of view, 3 mm slice thickness, and a 256 matrix) and axial T1-weighted spin-echo (repetition time of 500 ms, echo time of 20 ms, 22 cm field of view, 4 mm slice thickness with no interslice gap, and a 256 matrix). Axial and coronal T1-weighted spin-echo sequences were repeated, using a 256 or a 512 matrix, after an intravenous bolus injection of 0.1 mmol/kg gadolinium dimeglumine (Schering AG, Germany). In addition, a sagittal T1-weighted spin-echo and an axial T1 sequence with fat suppression (SPIR) were performed in four and 11 patients, respectively, following intravenous contrast. The MR and US results were assessed independently. Each study was assessed for: (1) site of the primary lesion; (2) multifocal thyroid disease; (3) extracapsular extension; (4) invasion of the trachea, q 2000 The Royal College of Radiologists
223
STAGING PAPILLARY CARCINOMA OF THE THYROID
Table 1 – Criteria for evaluation of US and MR images
Extracaspular extension Oesphageal invasion Tracheal invasion Major vessel invasion Metastatic lymphadenopathy
US criteria
MRI criteria
Poorly defined margin with heterogeneous echogenicity in adjacent fat or muscle Tumour invasion into the lumen Tumour invasion into the lumen Hypoechogenic wall and loss of mobility of tumour on swallowing Tumour invasion into the lumen Punctate calcification Cystic/ coagulation necrosis Extracapsular spread *Minimal axial diameter of 10 mm or more
Poorly defined margin with tumour extension into the adjacent fat or muscle Tumour invasion into the lumen Tumour invasion into the lumen Tumour encompassing more than 270 degrees of the carotid artery Tumour invasion into the lumen Calcification Cystic change/necrosis Extracapsular spread Minimal axial diameter of 10 mm or more
* Minimal axial diameter was chosen to correspond with the same criteria for assessing nodes on MRI.
oesophagus and major vessels; and (5) metastatic lymphadenopathy. The criteria for evaluation are shown in Table 1. For the purpose of scoring lymph-node metastases, each side of the neck was divided into the following regions: (1) submental and submandibular; (2) internal jugular; (3) supraclavicular fossa; (4) posterior triangle; (5) central nodes, defined as those in the anterior compartment of the neck that are removed en bloc with the thyroid and include the perithyroid, pre- and paratracheal and nodes around the recurrent laryngeal nerves. The results of imaging were correlated with pathological findings following surgery. A total thyroidectomy was performed in 13 patients and a hemithyroidectomy in one. Nodes outside the central compartment of the neck were resected on the basis of lymphadenopathy on clinical palapation, imaging or intra-operative assessment. A radical or functional neck dissection was performed in nine patients (bilateral in one). RESULTS
Tumour Site and Size The primary tumour ranged in size from 5 to 70 mm (mean 30 mm), of which six were situated in the right lobe and eight in the left. The site of the tumour was correctly identified by US in all cases. MRI failed to identify the 5-mm tumour.
Multifocal Disease Multiple nodules were present on pathological examination in seven patients of which five had benign multinodular thyroid and two had multifocal tumour. US identified all five cases of benign nodules and both cases of malignant multifocal tumour. MRI failed to identify one case of benign multinoduar thyroid and was unable to distinguish the two cases of malignant multifocal tumour from benign nodular disease.
spread into the muscle or fat. In the remaining three of nine cases, MRI correctly identified extracapsular tumour invasion into fat in one case that was missed on US, and both modalities missed two patients with early perithyroid extension into the fat. Invasion of the trachea was identified by MRI in one of the two (50%) patients and was missed by US in both cases. The one case of early oesophageal invasion was not identified by either modality.
Metastatic Cervical Lymphadenopathy Metastatic nodes were present at surgery in 19 nodal groups in 10 patients, involving the following groups; internal jugular chain (10), posterior triangle (four) supraclavicular fossa (one) and central group (four). MRI and US failed to identify four metastatic nodes in the central region in four patients. Three of these nodes were immediately adjacent to the thyroid and measured 0.2, 0.3 and 0.8 cm in maximum diameter; the fourth node was inferior to the thyroid in the paratracheal region and measured 0.6 cm. MRI and US correctly identified metastatic nodes in 14 of the 15 (93%) nodal groups outside the central group. In one patient, both MRI and US identified unenlarged nodes with no malignant features along the internal jugular chain, but failed to identify that two of the nine nodes (0.7 cm maximum diameter) were malignant. MRI revealed cystic areas of high signal intensity on both the T1- and T2-weighted sequences in eight of the 14 (57%) metastatic nodal groups and extracapsular spread in four of 14 (29%). US revealed that nodes in seven of the 14 (50%) metastatic nodal groups were hyperechoic with respect to muscle, nine (64%) contained cystic regions, four (29%) revealed extracapsular spread and 12 (86%) had punctate calcification (three nodes with punctate calcification were less than 1 cm in size, one of which was only 3 mm).
Extracapsular Extension
DISCUSSION
Extracapsular extension was present in nine of 14 patients (64%), involving the perithyroid muscle or fat in all nine, the trachea in two and the oesophagus in one patient. There was no invasion of the common carotid or internal jugular vein. MRI and US both identified six of the nine cases with extracapsular
Papillary carcinoma accounts for 60% of all thyroid malignancies. It is a well-differentiated tumour found especially in women, in the third or fourth decades of life. In general, patients with small, well-differentiated tumours confined to the thyroid, without lymph-node or distant metastases, can expect a
224
CLINICAL RADIOLOGY
long-term survival of nearly 100%. This figure falls with more advanced disease and overall the mortality rate is 11–16% of all cases [8]. In many centres, the extent of tumour is still assessed at the time of surgery, following which all macroscopic disease in the thyroid and lymph nodes is removed. However, pre-operative imaging is being used increasingly to alert the surgeon to extensive tumour spread which will require major reconstructive surgery, as well as identifying small volume disease that may be missed at surgery.
Multifocal Disease The decision as to whether to perform a total or partial thyroidectomy for papillary carcinoma is controversial. In many centres a total thyroidectomy is performed routinely. This is based on the finding that around 40% of papillary carcinomas are multifocal [9,10]. This multifocal tumour may be clinically occult in the contralateral lobe. However, it has been reported that following a hemithroidectomy for macroscopic disease restricted to one lobe, as assessed at the time of surgery, the eventual recurrence rate in the contralateral lobe will be only 5% [11]. For this reason some centres perform a hemithyroidectomy, so preserving normal thyroid tissue and reducing the risk of damage to the recurrent laryngeal nerve and parathyroid glands. In those centres performing limited thyroid resection surgery the pre-operative identification of disease in the contralateral gland will therefore influence the extent of surgery. In this study five patients had a multinodular goitre and two patients were found to have multifocal tumour. US was slightly better than MRI for identifying sites of abnormality in the thyroid and only US was able to distinguish tumour from a benign nodule in a multinodular goitre. This is because the US features of a papillary carcinoma are highly characteristic. These tumours are ill-defined and hypoechoic, and over 90% exhibit punctate calcification [12] (Fig. 1(a)). The calcification is due to the presence of psammoma bodies that arise when calcium is deposited onto the surface of the necrotic papillae [13]. There are, as yet, no reliable MR features of a thyroid nodule that differentiate benign from malignant disease [2–4] (Fig. 1(b)). Furthermore, in some cases of papillary carcinoma the thyroid gland may even appear normal [1], a finding that is supported in our study by the case in which a small 5 mm unifocal primary tumour could not be identified by MRI, even in retrospect.
(a)
Extracapsular Spread Extracapsular spread is present in 14% of patients with welldifferentiated carcinoma at presentation [14] and adversely affects prognosis. Both US and MRI were able to identify tumour confined within the thyroid tissue, but once it extended to the surface of the thyroid, MRI was marginally better in identification of extracapsular extension. One of the advantages of MRI was the better visualization of tumour invasion into the central region of the aerodigestive tract, an area where detail is obscured on US by the air in the trachea. US missed one case of tracheal invasion that was identified on MRI as a small tumour nodule that infiltrated the wall into the lumen. During this prospective study, a third patient with tracheal invasion was found to have disease detected only on MRI, but the results could not be included because the patient refused surgery. While frank
(b) Fig. 1 – (a) Ultrasound of the left lobe of the thyroid demonstrating a hypoechoic papillary carcinoma with punctate calcification (arrows). (b) Axial T1-weighted contrast-enhanced image of the patient in (a) demonstrating an enhancing papillary carcinoma in the left lobe (arrows).
invasion of the trachea could be identified (Fig. 2), MRI had difficulty, even in retrospect, in distinguishing early invasion from tumour abutting but not directly involving the wall of the trachea. Similar difficulty was found with assessment of invasion of the oesophagus.
STAGING PAPILLARY CARCINOMA OF THE THYROID
Fig. 2 – Axial T1-weighted contrast-enhanced MR image of the thyroid showing the primary tumour of the right lobe of the thyroid invading the posterior wall of the trachea (arrow).
The pre-operative identification of invasion into the aerodigestive tract has important implications for planning surgery. Although invasion of the aerodigestive tract only occurs in up to 6.5% of patients [8], failure to remove the tumour at this site increases the risk of recurrence [15]. Recurrence at this site can lead to uncontrolled asphyxia and haemorrhage into the airway, and is a major cause of mortality (38% of all deaths) [16]. In many centers, thyroid surgery is performed by general surgeons without expertise and knowledge of major reconstructive surgery. As a result, when unsuspected tumour invasion is encountered at surgery the tumour is ‘shaved off’ the trachea and the patient given radioactive iodide or external irradiation for the residual disease. This leads to an increased risk of recurrence [15]. The pre-operative identification of tumour invasion into the trachea or oesophagus therefore allows this small subgroup of patients to be referred to a surgeon with expertise in major reconstructive surgery, so improving long term local control and survival [8,15]. Invasion of the major vessels is rare in papillary carcinoma and was not present in this study.
225
are usually removed with the other anterior compartment nodes, but complete removal en bloc with the thyroid is technically difficult and requires meticulous dissection. Nodes at this site are important because they are along the route of lymphatic spread into the mediastinum. In this study, one small paratracheal metastatic node inferior to the thyroid was missed on imaging. However, from previous experience, MRI and US may identify nodes at this site and the detection of even small nodes in this region should probably be regarded with suspicion. The extent of surgical dissection of other nodal groups in the neck is controversial. The widely accepted practice is to avoid elective surgery and to reserve neck dissection for those cases in which lymphadenopathy is detected. In this study, 14 of the 15 (93%) nodal groups with metastatic involvement were correctly identified on both MRI and US. Both modalities failed to identify a patient with two small non-enlarged metastatic nodes along the internal jugular chain. Without extensive bilateral elective neck dissections in all patients or very long-term follow-up in this often indolent disease, it is not possible to determine the true accuracy of imaging in the detection of nodal metastases. However, this was not the aim of the study. Of note, there was one patient in this study who underwent bilateral neck dissection in two stages. In the first stage only an ipsilateral neck dissection was performed; the small contralateral metastatic nodes, detected at imaging, were not detected on intra-operative assessment. The contralateral internal jugular nodes were removed 1 month later and proven to be metastatic on histological evaluation. MRI revealed that over 50% of patients with abnormal nodes had cystic regions that were of high signal intensity on the T1 (Fig. 3) and T2-weighted images. This feature has been reported previously, with the cystic portion believed to represent thyroid protein (such as colloid or thyroglobulin), haemorrhage or tumour necrosis [1,7]. In the absence of cystic change diagnosis was mainly based on size or, in a minority of cases, on the presence of extracapsular spread. US was also able to demonstrate cystic change, in 64% of patients (Fig. 4), and extracapsular spread. Moreover, it had the added advantage of demonstrating two further features that are specific for metastases that could not be identified on MRI. The
Lymph-node Metastases Papillary carcinoma preferentially spreads via lymphatic rather than haematogenous routes. Lymph-node metastases are found in approximately 50% of all patients undergoing elective neck dissection [17] and this figure rises in patients with locally invasive tumours [18]. The lymphatics from the thyroid drain superiorly along the superior thyroid vessels to the precricoid and upper and mid-jugular nodes, laterally along the middle thyroid vein to the mid- and lower jugular nodes, and inferiorly along the inferior thyroid vessels to the pre- and paratracheal, and lower jugular nodes [19]. Less frequently, nodes in the posterior triangle, supraclavicular fossa, mediastinum and contralateral side of the neck may also be involved. Submandibular and submental lymph-node involvement is extremely rare. Both MRI and US failed to identify central metastatic nodes. These were mainly small nodes adjacent to the capsule. As this group of nodes is removed en bloc with the thyroid gland, the failure to identify these nodes did not influence management. Nodes inferior to the thyroid and above the suprasternal notch
Fig. 3 – Axial T1-weighted image of the thyroid. The primary tumour is in the left lobe/isthmus (arrow head). A large metastatic node demonstrating cystic change of high T1 signal intensity is present in the left internal jugular chain (arrows).
226
CLINICAL RADIOLOGY
of MRI is the identification of nodes in the upper mediastinum, which are not accessible to US. In conclusion, US should be the first investigation to be performed in staging papillary carcinoma because it is inexpensive and has the advantage over MRI of being able to identify small primary tumours, multifocal disease and lymph-node metastases. MRI should be reserved for those cases in which the tumour extends to the thyroid margin, so that extracapsular spread, especially into the trachea, can be assessed. It may also be used to assess the thyroid and nodes in the upper mediastinum when the inferior extent of disease can not be identified on US. REFERENCES
Fig. 4 – Ultrasound of the patient in Fig. 3 demonstrating a large, cystic lymph-node metastasis (arrows).
Fig. 5 – Axial US of the neck demonstrating a lymph-node metastasis from papillary carcinoma with punctate calcification (arrows) in the right internal jugular chain. The carotid artery is indicated by the curved arrow.
first was the increased echogenicity of metastatic nodes relative to muscle and the second was the presence of punctate calcification (Fig. 5). Three nodes were identified as metastatic only on the basis of punctate calcification and one of these was 3 mm in size. The identification of punctate calcification in nodes is a major advantage of US over MRI, especially as pathological studies show that lymph-node metastases from papillary carcinoma are frequently small and over 50% of all metastatic nodes are less than 3 mm in diameter [20]. Unfortunately, this advantage of US could not be reflected in the overall results of the study because of the small number of patients and the need to classify metastatic nodes into groups rather than individual nodes for the purpose of comparing the results with those from histological examination. One potential advantage
1 Som PM, Brandwein M, Lidov M, Lawson W, Biller HF. The varied presentations of papillary thyroid carcinoma cervical nodal disease: CT and MR findings. Am J Neuroradiol 1994;15:1129–1138. 2 Noma S, Kanaoka M, Minami S, et al. Thyroid masses: MR imaging and pathologic correlation. Radiology 1988;168:759–764. 3 Nakahara H, Noguchi S, Murakami N, et al. Gadolinium-enhanced MR imaging of thyroid and parathyroid masses. Head Neck Radiol 1997; 202:765–772. 4 Bagley JS, Ewen SWB, Smith FW, Krukowski ZH. Magnetic resonance imaging of thyroid swellings. Br J Surg 1996;83:828–829. 5 Auffermann W, Clark OH, Thurnher S, Galante M, Higgins CB. Recurrent thyroid carcinoma: characteristics on MR images. Radiology 1988;168:753–757. 6 Higgins CB, McNamara MT, Fisher MR. MR imaging of the thyroid. Am J Radiol 1986;147:1255–1261. 7 Takashima S, Sone S, Takayama F, et al. Papillary thyroid carcinoma: MR diagnosis of lymph node metastasis. Am J Neuroradiol 1997;19: 509–513. 8 Lawson W, Som ML, Biller HF. Papillary adenocarcinoma of the thyroid invading the upper air passage. Ann Otol 1977;86:751–755. 9 Clark RL, Ibanez ML, White EC. What constitutes an adequate operation for carcinoma of the thyroid. Arch Surg 1966;92:23–26. 10 Schindel J. Urgent extensive operation in cases of thyroid malignancy. Int Surg 1968;50:355–362. 11 Tollefsen HR, Shah JP, Huvos AG. Papillary carcinoma of the thyroid. Recurrence in the thyroid gland after initial surgical treatment. Am J Surg 1972;124:468–472. 12 Solbiati L, Livraghi T, Ballarati E, Lerace T, Crespi L. Thyroid gland. In: Solbiati L, Rizzatto G, eds. Ultrasound of Superficial Structures. London: Churchill Livingstone, 1995: 49–85. 13 Livolsi VA. Pathology of thyroid disease. In: Falk SA, ed. Thyroid Disease: Endocrinology, Surgery, Nuclear Medicine and Radiotherapy. Philadelphia: Lippincott-Raven, 1997: 65–104. 14 Silliphant WM, Klinck GH, Levitin MS. Thyroid carcinoma and death, a clinicopathological study of 193 autopsies. Cancer 1964;17:513–525. 15 Grillo HC, Zannini P. Resectional management of airway invasion by thyroid carcinoma. Ann Thorac Surg 1986;42:287–298. 16 Ishihara T, Yamakazi S, Kobayaski K, et al. Resection of the trachea infiltrated by thyroid carcinoma. Ann Surg 1982;195: 496. 17 Hutter RVP, Frazell EL, Foote FW. Elective radical neck dissection: an assessment of its use in papillary thyroid cancer. CA Cancer J Clin 1970;20:86–93. 18 Cody HS, Shah JP. Locally invasive, well-differentiated thyroid cancer. Am J Surg 1981;142:480–483. 19 Attie JN. Modified neck dissection in treatment of thyroid cancer: a safe procedure. Eur J Cancer Clin Oncol 1988;24:315–324. 20 Noguchi S, Noguchi A, Murakami N. Papillary carcinoma of the thyroid. Developing pattern of metastasis. Cancer 1970;26:1053–1060.
Staging Papillary Carcinoma of the Thyroid: Magnetic Resonance Imaging vs Ultrasound of the Neck A. D. KING, A. T . AHU JA, E . W . H. T O* , G. M . K. T SE† , C. MET R E W E L I Departments of Diagnostic Radiology & Organ Imaging, *Surgery and †Anatomical and Cellular Pathology, The Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, NT, Hong Kong Received: 18 April 1999 Revised: 27 July 1999
Accepted: 20 August 1999
AIM: To evaluate the role of magnetic resonance imaging (MRI) for staging local disease and lymph node metastases in papillary carcinoma of the thyroid by comparing MRI with ultrasound (US) of the neck. MATERIALS AND METHODS: Fourteen patients with papillary carcinoma underwent MRI and US. The images were prospectively reviewed for (1) identification of the primary lesion; (2) presence of multifocal thyroid disease; (3) extracapsular extension; (4) invasion into the trachea, oesophagus and major vessels; and (5) presence of lymph node metastases. Correlation was made with the surgical findings. RESULTS: The site of the primary lesion was correctly identified by US in 14 of 14 (100%) and by MRI in 13 of 14 (93%) of patients. Multifocal thyroid tumour was correctly identified by US in two of two patients (100%) and by MRI in zero of two (0%). Extracapsular extension (n ¼ 9), oesophageal invasion (n ¼ 1) and tracheal invasion (n ¼ 2) were identified by MRI in seven, zero, and one, and by US in six, zero and zero patients, respectively. Invasion of the major vessels was not seen. Metastatic cervical nodes were present in 19 nodal groups in 10 patients involving the internal jugular chain (n ¼ 10), posterior triangle (n ¼ 4) supraclavicular fossa (n ¼ 1) and central group (n ¼ 4). Both MRI and US failed to identify metastatic nodes in the central group but correctly identified 14 of the 15 nodal groups outside the central group. CONCLUSION: Ultrasound should be used as the first line of investigation for detecting the primary lesion, multifocal disease and cervical lymphadenopathy. In cases where the primary tumour is not surrounded by normal thyroid tissue MR imaging should be added to assess extracapsular spread, especially into the trachea. King, A. D. et al. (2000). Clinical Radiology 55, 222–226. q 2000 The Royal College of Radiologists Key words: MR imaging, ultrasound, thyroid, cancer.
Magnetic resonance imaging (MRI) is being used increasingly to image the thyroid gland [1–7] but as yet its role in diagnosis and management of the patient is unclear, especially when compared to the more established techniques for examining the thyroid, such as ultrasound (US). The aim of the study, therefore, was to compare MRI with US of the neck in the staging of papillary carcinoma of the thyroid. Papillary carcinoma was studied, not only because is it the most common thyroid malignancy, but also because imaging can influence the extent of surgical resection by identifying multifocal disease, extracapsular spread including invasion of the aerodigestive tract and regional lymph-node metastases. METHODS
Fourteen consecutive patients (six men, eight women; age range 16–81 years, mean 54 years) presenting with a thyroid Author for correspondence and guarantor of study: Dr A. D. King, Department of Diagnostic Radiology & Organ Imaging, Prince of Wales Hospital, Shatin, NT, Hong Kong. 0009-9260/00/030222+05 $35.00/0
mass confirmed to be a papillary carcinoma on fine needle aspiration (FNA) were entered into the study. Real-time ultrasound images were obtained using a 5–12 MHZ transducer (HDI 5000; Advanced Technology Laboratories, Bothell, WA). MR images were obtained on a 1.5 MRI unit (Philips Gyroscan, Best, Netherlands) using a neck coil (23 cm) and a surface coil. All patients underwent an axial dual turbo spin-echo (repetition time of 2500 ms, echo time of 35 ms, echo train length of 8, 23 cm field of view, 3 mm slice thickness, and a 256 matrix) and axial T1-weighted spin-echo (repetition time of 500 ms, echo time of 20 ms, 22 cm field of view, 4 mm slice thickness with no interslice gap, and a 256 matrix). Axial and coronal T1-weighted spin-echo sequences were repeated, using a 256 or a 512 matrix, after an intravenous bolus injection of 0.1 mmol/kg gadolinium dimeglumine (Schering AG, Germany). In addition, a sagittal T1-weighted spin-echo and an axial T1 sequence with fat suppression (SPIR) were performed in four and 11 patients, respectively, following intravenous contrast. The MR and US results were assessed independently. Each study was assessed for: (1) site of the primary lesion; (2) multifocal thyroid disease; (3) extracapsular extension; (4) invasion of the trachea, q 2000 The Royal College of Radiologists
223
STAGING PAPILLARY CARCINOMA OF THE THYROID
Table 1 – Criteria for evaluation of US and MR images
Extracaspular extension Oesphageal invasion Tracheal invasion Major vessel invasion Metastatic lymphadenopathy
US criteria
MRI criteria
Poorly defined margin with heterogeneous echogenicity in adjacent fat or muscle Tumour invasion into the lumen Tumour invasion into the lumen Hypoechogenic wall and loss of mobility of tumour on swallowing Tumour invasion into the lumen Punctate calcification Cystic/ coagulation necrosis Extracapsular spread *Minimal axial diameter of 10 mm or more
Poorly defined margin with tumour extension into the adjacent fat or muscle Tumour invasion into the lumen Tumour invasion into the lumen Tumour encompassing more than 270 degrees of the carotid artery Tumour invasion into the lumen Calcification Cystic change/necrosis Extracapsular spread Minimal axial diameter of 10 mm or more
* Minimal axial diameter was chosen to correspond with the same criteria for assessing nodes on MRI.
oesophagus and major vessels; and (5) metastatic lymphadenopathy. The criteria for evaluation are shown in Table 1. For the purpose of scoring lymph-node metastases, each side of the neck was divided into the following regions: (1) submental and submandibular; (2) internal jugular; (3) supraclavicular fossa; (4) posterior triangle; (5) central nodes, defined as those in the anterior compartment of the neck that are removed en bloc with the thyroid and include the perithyroid, pre- and paratracheal and nodes around the recurrent laryngeal nerves. The results of imaging were correlated with pathological findings following surgery. A total thyroidectomy was performed in 13 patients and a hemithyroidectomy in one. Nodes outside the central compartment of the neck were resected on the basis of lymphadenopathy on clinical palapation, imaging or intra-operative assessment. A radical or functional neck dissection was performed in nine patients (bilateral in one). RESULTS
Tumour Site and Size The primary tumour ranged in size from 5 to 70 mm (mean 30 mm), of which six were situated in the right lobe and eight in the left. The site of the tumour was correctly identified by US in all cases. MRI failed to identify the 5-mm tumour.
Multifocal Disease Multiple nodules were present on pathological examination in seven patients of which five had benign multinodular thyroid and two had multifocal tumour. US identified all five cases of benign nodules and both cases of malignant multifocal tumour. MRI failed to identify one case of benign multinoduar thyroid and was unable to distinguish the two cases of malignant multifocal tumour from benign nodular disease.
spread into the muscle or fat. In the remaining three of nine cases, MRI correctly identified extracapsular tumour invasion into fat in one case that was missed on US, and both modalities missed two patients with early perithyroid extension into the fat. Invasion of the trachea was identified by MRI in one of the two (50%) patients and was missed by US in both cases. The one case of early oesophageal invasion was not identified by either modality.
Metastatic Cervical Lymphadenopathy Metastatic nodes were present at surgery in 19 nodal groups in 10 patients, involving the following groups; internal jugular chain (10), posterior triangle (four) supraclavicular fossa (one) and central group (four). MRI and US failed to identify four metastatic nodes in the central region in four patients. Three of these nodes were immediately adjacent to the thyroid and measured 0.2, 0.3 and 0.8 cm in maximum diameter; the fourth node was inferior to the thyroid in the paratracheal region and measured 0.6 cm. MRI and US correctly identified metastatic nodes in 14 of the 15 (93%) nodal groups outside the central group. In one patient, both MRI and US identified unenlarged nodes with no malignant features along the internal jugular chain, but failed to identify that two of the nine nodes (0.7 cm maximum diameter) were malignant. MRI revealed cystic areas of high signal intensity on both the T1- and T2-weighted sequences in eight of the 14 (57%) metastatic nodal groups and extracapsular spread in four of 14 (29%). US revealed that nodes in seven of the 14 (50%) metastatic nodal groups were hyperechoic with respect to muscle, nine (64%) contained cystic regions, four (29%) revealed extracapsular spread and 12 (86%) had punctate calcification (three nodes with punctate calcification were less than 1 cm in size, one of which was only 3 mm).
Extracapsular Extension
DISCUSSION
Extracapsular extension was present in nine of 14 patients (64%), involving the perithyroid muscle or fat in all nine, the trachea in two and the oesophagus in one patient. There was no invasion of the common carotid or internal jugular vein. MRI and US both identified six of the nine cases with extracapsular
Papillary carcinoma accounts for 60% of all thyroid malignancies. It is a well-differentiated tumour found especially in women, in the third or fourth decades of life. In general, patients with small, well-differentiated tumours confined to the thyroid, without lymph-node or distant metastases, can expect a
224
CLINICAL RADIOLOGY
long-term survival of nearly 100%. This figure falls with more advanced disease and overall the mortality rate is 11–16% of all cases [8]. In many centres, the extent of tumour is still assessed at the time of surgery, following which all macroscopic disease in the thyroid and lymph nodes is removed. However, pre-operative imaging is being used increasingly to alert the surgeon to extensive tumour spread which will require major reconstructive surgery, as well as identifying small volume disease that may be missed at surgery.
Multifocal Disease The decision as to whether to perform a total or partial thyroidectomy for papillary carcinoma is controversial. In many centres a total thyroidectomy is performed routinely. This is based on the finding that around 40% of papillary carcinomas are multifocal [9,10]. This multifocal tumour may be clinically occult in the contralateral lobe. However, it has been reported that following a hemithroidectomy for macroscopic disease restricted to one lobe, as assessed at the time of surgery, the eventual recurrence rate in the contralateral lobe will be only 5% [11]. For this reason some centres perform a hemithyroidectomy, so preserving normal thyroid tissue and reducing the risk of damage to the recurrent laryngeal nerve and parathyroid glands. In those centres performing limited thyroid resection surgery the pre-operative identification of disease in the contralateral gland will therefore influence the extent of surgery. In this study five patients had a multinodular goitre and two patients were found to have multifocal tumour. US was slightly better than MRI for identifying sites of abnormality in the thyroid and only US was able to distinguish tumour from a benign nodule in a multinodular goitre. This is because the US features of a papillary carcinoma are highly characteristic. These tumours are ill-defined and hypoechoic, and over 90% exhibit punctate calcification [12] (Fig. 1(a)). The calcification is due to the presence of psammoma bodies that arise when calcium is deposited onto the surface of the necrotic papillae [13]. There are, as yet, no reliable MR features of a thyroid nodule that differentiate benign from malignant disease [2–4] (Fig. 1(b)). Furthermore, in some cases of papillary carcinoma the thyroid gland may even appear normal [1], a finding that is supported in our study by the case in which a small 5 mm unifocal primary tumour could not be identified by MRI, even in retrospect.
(a)
Extracapsular Spread Extracapsular spread is present in 14% of patients with welldifferentiated carcinoma at presentation [14] and adversely affects prognosis. Both US and MRI were able to identify tumour confined within the thyroid tissue, but once it extended to the surface of the thyroid, MRI was marginally better in identification of extracapsular extension. One of the advantages of MRI was the better visualization of tumour invasion into the central region of the aerodigestive tract, an area where detail is obscured on US by the air in the trachea. US missed one case of tracheal invasion that was identified on MRI as a small tumour nodule that infiltrated the wall into the lumen. During this prospective study, a third patient with tracheal invasion was found to have disease detected only on MRI, but the results could not be included because the patient refused surgery. While frank
(b) Fig. 1 – (a) Ultrasound of the left lobe of the thyroid demonstrating a hypoechoic papillary carcinoma with punctate calcification (arrows). (b) Axial T1-weighted contrast-enhanced image of the patient in (a) demonstrating an enhancing papillary carcinoma in the left lobe (arrows).
invasion of the trachea could be identified (Fig. 2), MRI had difficulty, even in retrospect, in distinguishing early invasion from tumour abutting but not directly involving the wall of the trachea. Similar difficulty was found with assessment of invasion of the oesophagus.
STAGING PAPILLARY CARCINOMA OF THE THYROID
Fig. 2 – Axial T1-weighted contrast-enhanced MR image of the thyroid showing the primary tumour of the right lobe of the thyroid invading the posterior wall of the trachea (arrow).
The pre-operative identification of invasion into the aerodigestive tract has important implications for planning surgery. Although invasion of the aerodigestive tract only occurs in up to 6.5% of patients [8], failure to remove the tumour at this site increases the risk of recurrence [15]. Recurrence at this site can lead to uncontrolled asphyxia and haemorrhage into the airway, and is a major cause of mortality (38% of all deaths) [16]. In many centers, thyroid surgery is performed by general surgeons without expertise and knowledge of major reconstructive surgery. As a result, when unsuspected tumour invasion is encountered at surgery the tumour is ‘shaved off’ the trachea and the patient given radioactive iodide or external irradiation for the residual disease. This leads to an increased risk of recurrence [15]. The pre-operative identification of tumour invasion into the trachea or oesophagus therefore allows this small subgroup of patients to be referred to a surgeon with expertise in major reconstructive surgery, so improving long term local control and survival [8,15]. Invasion of the major vessels is rare in papillary carcinoma and was not present in this study.
225
are usually removed with the other anterior compartment nodes, but complete removal en bloc with the thyroid is technically difficult and requires meticulous dissection. Nodes at this site are important because they are along the route of lymphatic spread into the mediastinum. In this study, one small paratracheal metastatic node inferior to the thyroid was missed on imaging. However, from previous experience, MRI and US may identify nodes at this site and the detection of even small nodes in this region should probably be regarded with suspicion. The extent of surgical dissection of other nodal groups in the neck is controversial. The widely accepted practice is to avoid elective surgery and to reserve neck dissection for those cases in which lymphadenopathy is detected. In this study, 14 of the 15 (93%) nodal groups with metastatic involvement were correctly identified on both MRI and US. Both modalities failed to identify a patient with two small non-enlarged metastatic nodes along the internal jugular chain. Without extensive bilateral elective neck dissections in all patients or very long-term follow-up in this often indolent disease, it is not possible to determine the true accuracy of imaging in the detection of nodal metastases. However, this was not the aim of the study. Of note, there was one patient in this study who underwent bilateral neck dissection in two stages. In the first stage only an ipsilateral neck dissection was performed; the small contralateral metastatic nodes, detected at imaging, were not detected on intra-operative assessment. The contralateral internal jugular nodes were removed 1 month later and proven to be metastatic on histological evaluation. MRI revealed that over 50% of patients with abnormal nodes had cystic regions that were of high signal intensity on the T1 (Fig. 3) and T2-weighted images. This feature has been reported previously, with the cystic portion believed to represent thyroid protein (such as colloid or thyroglobulin), haemorrhage or tumour necrosis [1,7]. In the absence of cystic change diagnosis was mainly based on size or, in a minority of cases, on the presence of extracapsular spread. US was also able to demonstrate cystic change, in 64% of patients (Fig. 4), and extracapsular spread. Moreover, it had the added advantage of demonstrating two further features that are specific for metastases that could not be identified on MRI. The
Lymph-node Metastases Papillary carcinoma preferentially spreads via lymphatic rather than haematogenous routes. Lymph-node metastases are found in approximately 50% of all patients undergoing elective neck dissection [17] and this figure rises in patients with locally invasive tumours [18]. The lymphatics from the thyroid drain superiorly along the superior thyroid vessels to the precricoid and upper and mid-jugular nodes, laterally along the middle thyroid vein to the mid- and lower jugular nodes, and inferiorly along the inferior thyroid vessels to the pre- and paratracheal, and lower jugular nodes [19]. Less frequently, nodes in the posterior triangle, supraclavicular fossa, mediastinum and contralateral side of the neck may also be involved. Submandibular and submental lymph-node involvement is extremely rare. Both MRI and US failed to identify central metastatic nodes. These were mainly small nodes adjacent to the capsule. As this group of nodes is removed en bloc with the thyroid gland, the failure to identify these nodes did not influence management. Nodes inferior to the thyroid and above the suprasternal notch
Fig. 3 – Axial T1-weighted image of the thyroid. The primary tumour is in the left lobe/isthmus (arrow head). A large metastatic node demonstrating cystic change of high T1 signal intensity is present in the left internal jugular chain (arrows).
226
CLINICAL RADIOLOGY
of MRI is the identification of nodes in the upper mediastinum, which are not accessible to US. In conclusion, US should be the first investigation to be performed in staging papillary carcinoma because it is inexpensive and has the advantage over MRI of being able to identify small primary tumours, multifocal disease and lymph-node metastases. MRI should be reserved for those cases in which the tumour extends to the thyroid margin, so that extracapsular spread, especially into the trachea, can be assessed. It may also be used to assess the thyroid and nodes in the upper mediastinum when the inferior extent of disease can not be identified on US. REFERENCES
Fig. 4 – Ultrasound of the patient in Fig. 3 demonstrating a large, cystic lymph-node metastasis (arrows).
Fig. 5 – Axial US of the neck demonstrating a lymph-node metastasis from papillary carcinoma with punctate calcification (arrows) in the right internal jugular chain. The carotid artery is indicated by the curved arrow.
first was the increased echogenicity of metastatic nodes relative to muscle and the second was the presence of punctate calcification (Fig. 5). Three nodes were identified as metastatic only on the basis of punctate calcification and one of these was 3 mm in size. The identification of punctate calcification in nodes is a major advantage of US over MRI, especially as pathological studies show that lymph-node metastases from papillary carcinoma are frequently small and over 50% of all metastatic nodes are less than 3 mm in diameter [20]. Unfortunately, this advantage of US could not be reflected in the overall results of the study because of the small number of patients and the need to classify metastatic nodes into groups rather than individual nodes for the purpose of comparing the results with those from histological examination. One potential advantage
1 Som PM, Brandwein M, Lidov M, Lawson W, Biller HF. The varied presentations of papillary thyroid carcinoma cervical nodal disease: CT and MR findings. Am J Neuroradiol 1994;15:1129–1138. 2 Noma S, Kanaoka M, Minami S, et al. Thyroid masses: MR imaging and pathologic correlation. Radiology 1988;168:759–764. 3 Nakahara H, Noguchi S, Murakami N, et al. Gadolinium-enhanced MR imaging of thyroid and parathyroid masses. Head Neck Radiol 1997; 202:765–772. 4 Bagley JS, Ewen SWB, Smith FW, Krukowski ZH. Magnetic resonance imaging of thyroid swellings. Br J Surg 1996;83:828–829. 5 Auffermann W, Clark OH, Thurnher S, Galante M, Higgins CB. Recurrent thyroid carcinoma: characteristics on MR images. Radiology 1988;168:753–757. 6 Higgins CB, McNamara MT, Fisher MR. MR imaging of the thyroid. Am J Radiol 1986;147:1255–1261. 7 Takashima S, Sone S, Takayama F, et al. Papillary thyroid carcinoma: MR diagnosis of lymph node metastasis. Am J Neuroradiol 1997;19: 509–513. 8 Lawson W, Som ML, Biller HF. Papillary adenocarcinoma of the thyroid invading the upper air passage. Ann Otol 1977;86:751–755. 9 Clark RL, Ibanez ML, White EC. What constitutes an adequate operation for carcinoma of the thyroid. Arch Surg 1966;92:23–26. 10 Schindel J. Urgent extensive operation in cases of thyroid malignancy. Int Surg 1968;50:355–362. 11 Tollefsen HR, Shah JP, Huvos AG. Papillary carcinoma of the thyroid. Recurrence in the thyroid gland after initial surgical treatment. Am J Surg 1972;124:468–472. 12 Solbiati L, Livraghi T, Ballarati E, Lerace T, Crespi L. Thyroid gland. In: Solbiati L, Rizzatto G, eds. Ultrasound of Superficial Structures. London: Churchill Livingstone, 1995: 49–85. 13 Livolsi VA. Pathology of thyroid disease. In: Falk SA, ed. Thyroid Disease: Endocrinology, Surgery, Nuclear Medicine and Radiotherapy. Philadelphia: Lippincott-Raven, 1997: 65–104. 14 Silliphant WM, Klinck GH, Levitin MS. Thyroid carcinoma and death, a clinicopathological study of 193 autopsies. Cancer 1964;17:513–525. 15 Grillo HC, Zannini P. Resectional management of airway invasion by thyroid carcinoma. Ann Thorac Surg 1986;42:287–298. 16 Ishihara T, Yamakazi S, Kobayaski K, et al. Resection of the trachea infiltrated by thyroid carcinoma. Ann Surg 1982;195: 496. 17 Hutter RVP, Frazell EL, Foote FW. Elective radical neck dissection: an assessment of its use in papillary thyroid cancer. CA Cancer J Clin 1970;20:86–93. 18 Cody HS, Shah JP. Locally invasive, well-differentiated thyroid cancer. Am J Surg 1981;142:480–483. 19 Attie JN. Modified neck dissection in treatment of thyroid cancer: a safe procedure. Eur J Cancer Clin Oncol 1988;24:315–324. 20 Noguchi S, Noguchi A, Murakami N. Papillary carcinoma of the thyroid. Developing pattern of metastasis. Cancer 1970;26:1053–1060.