THYROIDVolume 11, Number 12, 2001Mary Ann Liebert, Inc.

Prognostic Significance of Magnetic Resonance Findings in Advanced Papillary Thyroid Cancer

Shodayu Takashima,1 Tsuyoshi Matsushita,1 Fumiyoshi Takayama,1 Masumi Kadoya,1 Minoru Fujimori,2

and Tetsuro Kobayashi3

We assessed the prognostic importance of magnetic resonance (MR) findings in locally advanced papillary thy-roid cancer. MR findings, clinical data, and pathologic (and surgical) data for 66 patients, including 51 womenand 15 men with a mean age of 57 years, who had primary surgery for papillary thyroid cancers were corre-lated with prognosis. Mean follow-up was 27.5 months (range, 5–117 months). Recurrence was seen in 18 pa-tients (27%). In univariate analyses, age of 60 years or more (p 5 0.0066), male gender (p 5 0.0373), six MR find-ings (tumor size of $4 cm ([p 5 0.0002], ill-defined margins ([p, 0.0001], tumor extension of the trachea [p 50.0337], carotoid artery [p 5 0.0028]), esophagus [p , 0.0001], and lymph nodes [p 5 0.0005]), and three patho-logic findings (tumor extension of soft tissues [p 5 0.0288], carotid artery [p 5 0.0013], and esophagus [p, 0.0001])had a significant adverse effect on disease-free survival. In multivariate analyses, tumor size (p 5 0.0169) andnodal metastasis (p 5 0.0393) determined on MR imaging and pathologic esophageal invasion (p 5 0.0016) werethe only significant independent variables. Esophageal invasion was accurately diagnosed with MR imaging(94% accuracy). MR findings may contain prognostic importance of locally advanced papillary thyroid cancer.

1153

Introduction

PAPILLARY THYROID CANCER is the most common malignantneoplasm in the thyroid gland and account for approxi-

mately 60%–75% of all thyroid cancer (1–3). This tumor peaksin the third or fourth decades of life and predominantly af-fects women with a female to male ratio of 1.8:1 to 3.1:1 (1–5).Patients with tumor have good prognosis and 10-year sur-vival rates exceed 90% (2,3). Until now, several clinical andpathologic factors have been reported to influence progno-sis of patients with this neoplasm. Such prognostic parame-ters include distant metastases, extrathyroidal tumor exten-sion, size of the primary lesion, age of the patient, anddifferentiation of the tumors. Distant metastases, direct in-vasion of the primary tumor into the surrounding structures,papillary cancers larger than 1.5 cm in diameter, age over 40at the time of diagnosis, and poorly differentiated tumors in-dicate poor prognosis (2–7). The prognostic importance ofregional lymph node metastasis is debatable (2–9). However,many investigators insisted that the nodal metastasis had anadverse effect on survival or recurrence rate (2,6–9).

There are many reports on magnetic resonance (MR) imag-ing appearances in focal and diffuse thyroid diseases (10–12).There are many reports in which multivariate regression

analysis has been carried out to clarify the independent prog-nostic factors in papillary thyroid cancer. However, all of theprevious researchers focused on clinical, pathologic, biolog-ical, or therapeutic aspects (3,5–7,13–15). To our knowledge,there is no literature in which MR imaging findings werecorrelated with prognosis of patients with the tumor. Thisstudy was designed to investigate if MR findings have in-dependent prognostic significance in a high-risk tumorgroup of papillary thyroid cancer.

Materials and Methods

This retrospective study consisted of 66 consecutive pa-tients (51 females, 15 males) in one institution, aged 20–85years (mean, 57 years), who underwent MR imaging to de-fine the extent of the thyroid tumors or lymph node metas-tases before their primary surgical treatment for papillarythyroid cancer between 1989 and 1999. One to 26 dayselapsed between MR imaging and surgery. The diagnosis ofpapillary thyroid cancer was obtained by fine-needle aspi-ration biopsy (FNAB) either with guidance of palpation (n 560) or ultrasound (US) (n 5 6). In our hospital, palpable thy-roid masses initially underwent palpation-guided FNAB andUS-guided FNAB was performed in the patients in whom

Departments of 1Radiology and 2Surgery, Shinshu University School of Medicine, Matsumoto, Japan.3Department of Surgery, Ikeda Municipal Hospital, Ikeda, Japan.

insufficient materials were obtained by palpation-guidedFNAB or discrepancy between clinical diagnosis and palpa-tion-guided FNAB diagnosis was found. For the primarystaging, US of the neck, bone scans using technetium 99mpertechnetate scintigraphy, computed tomography (CT) ofthe chest, and US or CT of the abdomen were performed inall patients. In addition, US-guided FNAB was performedfor primary staging in another coexistent nodules in the thy-roid gland in seven patients or in the lymph nodes in fivepatients. The patients who had had distant metastases at thetime of diagnosis were excluded from this study.

MR imaging was performed with a Magnetom (Siemens,Erlangen, Germany) or a Signa unit (GE Medical Systems,Milwaukee, WI) using a volume neck coil (Medical Ad-vances, Milwaukee, WI) or a Helmholtz coil with a 256 3192 acquisition matrix. Conventional nonenhanced spin-echo T1-weighted images (T1WI) (700/13 or 900/15) [repe-tition time ms/echo time ms]) were obtained in all patients.T2-weighted images (T2WI) were obtained with conven-tional spin-echo sequence (2,000/70) in 28 patients and witha fast spin-echo T2WI (3,200/91) in 38 patients. Of the 66 pa-tients, contrast-enhanced T1WI (700/13 or 900/15) were ob-tained in 55 patients immediately after intravenous bolus in-jection of 0.1 or 0.2 mmol/kg gadopentate dimeglumine(Magnevist; Nihon Shering, Suita, Japan). The number of ex-citations was two for T1WI and one or two for T2WI. A to-tal of 18 axial images from the level of the mandibular an-gle to the sternal notch were obtained with a slice thicknessof 5 mm and an intersection gap of 1.0–2.0 mm using pre-saturation pulses placed both inferior and superior to the ac-quired slices.

Without knowledge of the pathologic and surgical find-ings, two head and neck radiologists (F.T., S.T.) indepen-dently evaluated MR images for T2-weighted signal inten-sity and margin characteristics of the primary tumors of thethyroid gland, and presence or absence of tumor invasion ofthe trachea, esophagus, and the regional cervical lymphnodes based on a two-point scale method. T2-weighted sig-nal intensity included hypointensity to the thyroid gland ornot (isointensity or hyperintensity) and margin characteris-tics included (partially or totally) ill-defined margins andwell-defined margins. T2-signal characteristics were evalu-ated for the signal intensity of solid portions with no ap-parent calcification in the tumors. When the tumor had dif-ferent T2-signals, signal intensity in predominant areas ofthe tumor was used for evaluation. When the primary tu-mor occurred in multiple form, the largest lesion was usedfor evaluation. The degrees of tumor circumference of thecommon carotid artery were graded into four consecutiveclasses; class 1, ,90 degrees; class 2, $90 degrees and ,180degrees; class 3, $180 degrees and ,270 degrees; class 4,$270 degrees.

MR features used for the prediction of tumor invasion inour series were T2-high–signal areas in the tracheal cartilageor intraluminal mass for the trachea, erosion of theesophageal wall by the tumor for the esophagus, degrees oftumor circumferences of $270 degrees or more for the com-mon carotid artery (16), and 8 mm or more in the minimalaxial diameter or a cystic node for the regional lymph nodes(17). k statistics for interobserver agreement for six MR imag-ing findings were 0.68–0.83, which indicated almost perfector substantial agreement (18). In the current study, we used

consensus method to examine prognostic statistics based onone final set of the results. One of the radiologists (S.T.) alsomeasured the greatest axial diameters of the primary tumorson MR images.

Surgical resection of the thyroid tumors consisted of lobec-tomy in 4 patients (6%), subtotal thyroidectomy in 33 (50%),and total thyroidectomy in 29 (44%). The rate of total thy-roidectomy was low in our institute because surgeons pre-ferred to preserve the recurrent laryngeal nerve and at leastone parathyroid gland and to avoid myxedema, which couldbe incurred, when older patients with total thyroidectomyfailed to take thyroid hormone. Modified neck dissection wasperformed in patients with palpable lymph nodes or meta-static nodes diagnosed with MR imaging. In patients withclinically and radiologically negative neck, a dissection of theregional central lymph nodes was performed; when meta-static nodes were identified at the frozen section studies,modified neck dissection was completed. We retrospectivelyreviewed pathologic and surgical data regarding tumor in-volvement of the regional lymph nodes and adjacent struc-tures; surgical diagnosis of tumor invasion of the surround-ing structures was obtained when the tumor margins werepositive and residual tumor was apparent in the corre-sponding organs and pathologic diagnosis was obtainedwhen pathologic proof of tumor invasion was verified in sur-gically resected specimens.

We then performed Kaplan-Meier method for disease-freesurvival using each factor of MR findings, and pathologicand surgical diagnoses. The patients were classified into twoage groups ($60 years or ,60 years). Tumor diameters werecategorized into two groups (,4 cm or $4 cm in greatest ax-ial diameters). Follow-up periods for the patients rangedfrom 5 to 117 months (mean 27.5 months). The log-rank testwas used for calculating statistical differences.

Next, we performed multivariate Cox proportional haz-ards regression for disease-free survival to know the signif-icant independent prognostic factors. The modeling was con-ducted in a step-up manner. The selection was stopped whenno other variables added to the model proved significant atthe 0.05 level. p values for the models were calculated withthe likelihood ratio. These analyses were conducted for thethree groups of independent variables; one group of all theMR findings (T2 signal, margin characteristics, tumor diam-eters, and MR criteria for tumor invasion of the lymph node,trachea, esophagus, and the carotid artery); another groupof all the pathologic and surgical data (tumor invasion of thelymph node, extrathyroid soft tissues, trachea, esophagus,and the carotid artery); the other group of all the indepen-dent variables. A p value of less than 0.05 was considered tobe significant. All of the statistical calculations were per-formed with SPSS software (SPSS, Chicago, Ill).

Results

MR findings versus pathologic and surgical data

The maximum axial diameter of the primary thyroid tu-mors was 3.2 6 1.6 cm (mean 6 standard deviation [SD]).According to the TNM classification of UICC (19), the pri-mary tumors were classified as T2 in 18 patients, T3 in 2 pa-tients, and T4 in 46 patients.

All tumors were detected with one or combinations of thethree MR pulse sequences. On T2WI, 10 tumors (15%) ap-

TAKASHIMA ET AL.1154

peared hypointense and the other 56 tumors (85%) showedisointensity (n 5 10; 15%) or hyperintensity (n 5 46; 70%). Ill-defined margins were seen in 21 tumors (32%) and well-de-fined margins in 45 tumors (68%). In pathologic studies, atotally encapsulated primary tumor was found in 4 (6%) ofthe 66 tumors and all 4 tumors showed well-defined mar-gins on MR images. A partial or absent capsule was seen inthe other 62 tumors (94%).

On MR images, lymph nodes of 8 mm or more in min-imal axial diameter or a cystic node were discovered in 29(44%, Figs. 1A and 1B) of the 66 cases, T2-high–signal ar-eas in the tracheal cartilage or intraluminal mass in 28cases (42%, Figs. 2A, 2B, and 2C), and erosion of theesophageal wall by the tumor in 21 cases (32%, Figs. 3A,3B, and 3C). Although cystic nodes were seen in 14 pa-tients, all these patients had 8 mm or larger lymph nodes.

PROGNOSTIC SIGNIFICANCE OF MRI IN THYROID CANCER 1155

FIG. 1. Primary papillary cancer with lymph node metastasis in a 57-year-old woman. Primary tumor (arrowheads) of$4 cm in the right lobe and the isthmus of the thyroid gland with well-defined margins appears as isointense areas on T1-weighted magnetic resonance (MR) images (A) and hyperintense areas on T2-weighted MR images (B). Multiple metasta-tic nodes (asterisks) in the ipsilateral neck were seen. Some of the nodes measured $8 mm in minimal axial diameter andsome of them developed extensive cystic change that appears hyperintense both on T1-weighted and T2-weighted MR im-ages.

A B

A

C

B

FIG. 2. Papillary thyroid cancer with tracheal invasion ina 35-year-old man. Primary tumor (white and black arrow-heads) of the greatest axial diameter of $4 cm in the left lobeand the isthmus of the thyroid gland with ill-defined mar-gins appears as hypointense areas on T1-weighted magneticresonance (MR) images (A) and inhomogeneous hyperin-tense areas on both T2-weighted MR images (B) and fat-sup-pressed contrast-enhanced T1-weighted MR images (C). Tu-mor extension into the tracheal lumen (arrow) andsurrounding soft tissues is apparent. Tumor invasion of thetrachea was diagnosed pathologically.

Tumor circumferences of the carotid artery were estimatedas class 1 in 43 cases (65%), class 2 in 11 (17%), class 3 in8 (12%), and class 4 in 4 (6%). Thus, carotid invasion waspredicted with MR imaging in the four cases (Figs.3A–3C). Diagnostic statistics of these MR criteria, whenusing the pathologic and surgical data as a gold standard,are summarized in Table 1. Sensitivity, specificity, and ac-curacy of MR criteria for tumor involvement were 60%,

95%, and 70%, respectively, for the lymph node; 95%, 84%,and 88%, respectively, for the trachea; 40%, 100%, and91%, respectively, for the carotid artery; and 100%, 92%,and 94%, respectively, for the esophagus. p values for thecorrelation between MR criteria and pathologic and sur-gical diagnoses for tumor invasion that were calculatedwith Fisher’s exact tests were #0.001 for all of the fourstructures.

TAKASHIMA ET AL.1156

A

C

B

FIG. 3. Papillary thyroid cancer with carotid andesophageal invasion in a 65-year-old woman. Primary tu-mor (white and black arrowheads) of $4 cm in the left lobeand the isthmus of the thyroid gland with ill-defined mar-gins appears as inhomogeneous hyperintense areas on bothT1-weighted magnetic resonance (MR) images (A) and T2-weighted MR images (B). Erosion (arrow) of the esophagealwall is apparent. Tumor circumference of the commoncarotid artery (asterisk) was estimated as $270°. The tumor(white and black arrowheads) appears hypoechoic on trans-verse ultrasonography (C). Evaluation of tumor invasion tothe trachea and esophagus with ultrasonography was hin-dered by the air in the tracheal lumen and acoustic shadowscaused by dense calcification (arrows) in the tumor. Patho-logic and surgical diagnosis of tumor invasion of the esoph-agus and the carotid artery was established. T, trachea.

TABLE 1. DIAGNOSTIC STATISTICS OF MR IMAGING CRITERIA FOR DIAGNOSIS OF TUMOR INVOLVEMENT

MR criteria of tumor Sensitivity Specificity Accuracyinvolvement (%) (%) (%) Positive Negative

Lymph node$8 mm or a cystic node 60 (28/47) 95 (18/19) 70 (46/66) 97 (28/29) 49 (18/37)

TracheaT2-high signal or 95 (21/22) 84 (37/44) 88 (58/66) 75 (21/28) 97 (37/38)

intraluminal massCarotid arteryTumor circumference 40 (4/10) 100 (56/56) 91 (60/66) 100 (4/4) 90 (56/62)

of $270°EsophagusErosion by tumor 100 (17/17) 92 (45/49) 94 (62/66) 81 (17/21) 100 (45/45)

Numbers in parentheses indicate exact numbers of patients.MR, magnetic resonance.

Predictive value (%)

Statistical results of prognosis

Recurrent tumors were identified in 18 (27%) of the 66 pa-tients. Recurrence was seen in the lymph nodes in 6 cases,lymph nodes and distant organs in 4, thyroid bed and lymphnodes and distant organs in 4, thyroid bed in 2, and distantorgans in 2. Recurrent tumor in the lymph nodes was diag-nosed when the tumor was found either in the areas thatwere successfully eradicated by neck dissections or in the ar-eas in which no tumor had been detected on the first radio-logic follow-up studies after surgery. Metastasized distantorgans included lungs (n 5 7), bone (n 5 3), soft tissues (n 52), brain (n 5 2), and liver (n 5 1). Tumor invasion waspathologically and surgically diagnosed for the lymph nodein 47 cases (71%), the extrathyroid soft tissues in 46 (70%),the trachea in 22 (33%), the carotid artery in 10 (15%), andthe esophagus in 17 (26%).

Kaplan-Meier method showed that the age of 60 years ormore (p 5 0.0066) and male gender (p5 0.0373) significantlyand adversely affected disease-free survival (Table 2). Of MRfindings, a statistically significant difference was noted for thetumor size (p5 0.0002), margin characteristics (p , 0.0001),and MR criteria for tumor involvement to the trachea (p 50.0337), carotid artery (p5 0.0028), esophagus (p , 0.0001),and the lymph node (p5 0.0005). Large tumors, ill-defined

margins, and MR diagnosis of tumor invasion of these fourorgans inversely related to disease-free survival. Of the patho-logic and surgical data, a statistically significant difference wasseen in tumor invasion of the extrathyroid soft tissues (p 50.0288), carotid (p5 0.0013) and the esophagus (p , 0.0001)that adversely influenced disease-free survival.

With Cox multivariate regression analysis, we obtained asignificant result for all of the three models (p , 0.0001 forthree groups of MR data, pathologic and surgical data, andtotal data; Table 3). Of the MR findings, tumor diameters(p 5 0.0088) and MR prediction of tumor involvement of theesophagus (p 5 0.0014) and the lymph node (p 5 0.0476)were significant independent prognostic indicators. Of thepathologic and surgical data, esophageal invasion was theonly significant adverse prognostic factor (p , 0.0001). Whenthe model of all the parameters was assessed, tumor size of4 cm or more (p 5 0.0169) and lymph node metastasis (p 50.0393) on MR imaging and pathologic esophageal invasion(p 5 0.0016) were significant independent prognostic factorsthat related adversely to disease-free survival.

Discussion

It must be acknowledged that the study population wasbiased toward patients with poor prognosis, because this se-

PROGNOSTIC SIGNIFICANCE OF MRI IN THYROID CANCER 1157

TABLE 2. RESULTS OF KAPLAN-MEIER METHOD FOR DISEASE-FREE SURVIVAL IN 66 PATIENTS

Independent variables p Adverse factor

Age 0.0066 $60 yearsGender 0.0373 MaleMR imaging findingsMaximum diameter 0.0002 Large sizeMargin characteristics ,.0001 III definitionT2-signal of tumors 0.8199 —T2-high signal of cartilage or intraluminal mass 0.0337 PresenceTumor circumference of carotid artery of $270° 0.0028 PresenceEsophageal erosion ,.0001 Presence$8 mm lymph node or a cystic node 0.0005 Presence

Pathologic and surgical dataLymph node metastasis 0.1643 —Extrarthyroid extension 0.0288 PresenceTracheal invasion 0.0652 —Carotid invasion 0.0013 PresenceEsophageal invasion ,.0001 Presence

p values were calculated with log-rank tests.MR, magnetic resonance.

TABLE 3. RESULTS OF STEP-UP COX PROPORTIONAL-HAZARDS MODELING FOR DISEASE-FREE SURVIVAL IN 66 PATIENTS

Group of independent p for the Selected significantvariables model independent variables p Adverse factor Relative risk

MR imaging findings ,0.0001 Maximum diameter ,0.0088 Large size 4.746Esophageal erosion ,0.0014 Presence 0.171$8 mm lymph node or a cystic node ,0.0476 Presence 0.214

Pathological and surgical data ,0.0001 Esophageal invasion ,.0001 Presence 0.120All of the parameters ,0.0001 Maximum diameter ,0.0169 Large size 4.136

Esophageal invasion ,0.0016 Presence 0.189$8 mm lymph node or a cystic node ,0.0393 Presence 0.201

MR, magnetic resonance.

ries included only those patients who were clinically sus-pected of having invasive tumors or nodal metastasis. Themean age (57 years) in our patients was much higher thanthose reported in the literature (1–4). Papillary thyroid can-cers, when occurring in elderly patients, tend to be more ag-gressive and invade the surrounding structures with resul-tant poor prognosis (4,7). Actually, the mean tumor size (3.2cm), relapse rate (27%), and the incidence of extrathyroid tu-mor extension (70%), and tumor invasion to the surround-ing organs such as the trachea (33%), carotid artery (15%),and the esophagus (26%) were greater than those docu-mented in the literature (1–4,20–22). However, on the con-trary, such patient population enabled us to easily evaluatethe prognostic factors in relatively short follow-up periods.

In univariate analyses of the current series, among the clin-ical and pathologic findings, age of 60 years or more, malegender extrathyroid tumor spread, and carotid or esophagealinvasion had a significant inverse effect on disease-free sur-vival. The first three factors were similarly stated in the lit-erature (2–7,13). However, there is no literature in which thespecific organs invaded by the tumor greatly influence theprognosis of patients with papillary thyroid cancer. Ourstudy revealed that tumor invasion of the esophagus (p ,0.0001) and carotid artery (p 5 0.0013) was of more prog-nostic importance than the mere tumor spread to the ex-trathyroid soft tissues (p 5 0.0288).

Of the MR findings, large tumor ($4 cm), ill-defined mar-gins and tumor invasion to the trachea, carotid artery, esoph-agus, and the lymph node were the significant prognostic in-dicators in univariate analyses. Tumor size is a generallyaccepted prognostic factor (2,4,6,13). In our pathologic stud-ies, well-defined margins were not a reliable sign for totalencapsulation of thyroid tumors but ill-defined margins pre-cluded a total encapsulation. We think that the ill definitionindicates macroscopic tumor invasion to the surrounding tis-sues. This fact may explain the reason for poor prognosis ofill-defined margins. The MR criteria for tumor invasion ofthe carotid artery and the nodal metastasis used in the studyhad been proposed by other authorities (16,17). The MR cri-teria for esophageal or tracheal invasion were not docu-mented in the literature. However, we think that the MR fea-tures may strongly reflect pathologic tumor invasion of suchorgans, because an intimate significant correlation (p ,0.001) between those findings and pathologic or surgical di-agnoses was found.

Interestingly, the adverse prognostic factor on disease-freesurvival was the swollen nodes on MR imaging but notpathologic nodal metastasis. Our results agree with the opin-ion that lymph node metastasis which significantly raises therecurrence rate and mortality was not microscopic metasta-sis but macroscopic metastasis in the lymph nodes (2,7,9). Aspreviously proposed (17), we suggest that 8 mm or greaterin the minimal axial diameter or a cystic node irrelevant toits size is most accurate for predicting metastatic lymphnodes from papillary thyroid cancer. Although it was notsensitive (60% sensitivity) because of the high incidence ofmicroscopic metastasis, the MR criterion for nodal metasta-sis was highly specific (95% specificity).

In multivariate analyses, tumor size and nodal metastasison MR imaging and pathologic esophageal invasion werethe only significant adverse independent factors in ourstudy. Of these, the most important prognostic factor was

esophageal invasion (p 5 0.0016), followed by tumor size(p 5 0.0169), and by nodal metastasis (p 5 0.0393).Esophageal invasion was accurately predicted with our MRcriterion.

In our study, tumor infiltration of the adjacent organs suchas the trachea (88% accuracy), esophagus (94% accuracy), orthe carotid artery (91% accuracy) was predicted accuratelywith MR imaging, probably because of the superior soft tis-sue contrast of this radiologic modality. We think that de-termination of presence or absence of tumor invasion to thesurrounding organs with MR imaging will be clinically ben-eficial not only for the proper treatment planning but alsofor prediction of correct prognosis. Aggressive local resec-tion may be warranted for the patients with tumor invasionto the adjacent tissues, because gross removal of the tumorscan improve the disease-free survival (20,21).

In conclusion, MR findings may contain prognostic infor-mation of locally advanced papillary thyroid cancer. How-ever, the number of patients was small in this series, so morework in a large group of patients with long-term follow-upis necessary before a final conclusion can be made regard-ing the prognostic importance of MR imaging.

References

1. Woolner LB, Beahrs OH, Black BM, McConahey WM, Keat-ing FR Jr 1961 Classification and prognosis of thyroid carci-noma: A study of 885 cases observed in a thirty year period.Am J Surg 102:354–387.

2. Mazzaferri EL 1987 Papillary thyroid carcinoma: Factors in-fluencing prognosis and current therapy. Semin Oncol14:315–332.

3. Gilliland FD, Hunt WC, Morris DM, Key CR 1997 Prognos-tic factors for thyroid carcinoma. A population-based studyof 15,698 cases from the surveillance, epidemiology and endresults (SEER) program 1973–1991. Cancer 79:564–573.

4. Carcangiu ML, Zampi G, Pupi A, Castagnoli A, Rosai J 1985Papillary carcinoma of the thyroid: A clinicopathologicstudy of 241 cases treated at the University of Florence, Italy.Cancer 55:805–828.

5. Lorentz TG, Lau PWK, Lo CY, Law WL, Wan KY, Lauder IJ1994 Multivariate analysis of risk factors influencing sur-vival in 110 ethnic Chinese with papillary thyroid cancer.World J Surg 18:547–551.

6. Simpson WJ, McKinney SE, Carruthers JS, GospodarowiczMK, Sutcliffe SB, Panzarrella T 1987 Papillary and follicularthyroid cancer. Prognostic factors in 1,578 patients. Am JMed 83:479–488.

7. Tubiana M, Schlumberger M, Rougier P, Laplauche A, Ben-hamou E, Gardet P, Caillon B, Travagli JP, Parmentier C 1985Long-term results and prognostic factors in patients withdifferentiated thyroid carcinoma. Cancer 55:794–804.

8. Haywood J, Clark OH, Dunphy JE 1978 Significance oflymph node metastasis in differentiated thyroid cancer. AmJ Surg 136:107–112.

9. Sellers M, Beenken S, Blankenship A, Soong SJ, Turbat-Her-rena E, Urist M, Maddox W 1992 Prognostic significance ofcervical lymph node metastases in differentiated thyroidcancer. Am J Surg 164:578–581.

10. Higgins CB, McNamara MT, Fisher MR, Clark OH 1986 MRimaging of the thyroid. AJR 147:1255–1261.

11. Gefter WB, Spritzer CE, Eisenberg B, LiVolsi VA, Axel L,Velchik M, Alavi A, Schenck J, Kressel HY 1987 Thyroid

TAKASHIMA ET AL.1158

imaging with high-field-strength surface-coil MR. Radiology164:483–490.

12. Takashima S, Nomura N, Noguchi Y, Matsuzuka F, Inoue T1995 Primary thyroid lymphoma: Evaluation with US, CT,and MRI. J Comput Assist Tomogr 19:282–288.

13. Hay ID, Bergstralh EJ, Goellner JR, Grant CS 1993 Predict-ing outcome in papillary thyroid carcinoma: Developmentof a reliable prognostic scoring system in a cohort of 1779patients surgically treated at one institution during1940–1989. Surgery 114:1050–1057.

14. Akslen LA 1993 Prognostic importance of histologic grad-ing in papillary thyroid carcinoma. Cancer 72:2680–2685.

15. Pasieka JL, Zedenius J, Auer G, Grimelins L, Hoog A, Lun-dell G, Wallin G, Backsahl M 1992 Addition of nuclear DNAcontent to the AMES risk-group classification for papillarythyroid cancer. Surgery 112:1154–1159.

16. Yousem DM, Hatabu H, Hurst RW, Seigerman HM, Mon-tone KT, Weinstein GS, Hayden RE, Goldberg AN, BigelowDC, Kotapka MJ 1995 Carotid artery invasion by head andneck masses: Prediction with MR imaging. Radiology195:715–720.

17. Takashima S, Sone S, Takayama F, Wang Q, Kobayashi T,Horii A, Yoshida J 1998 Papillary thyroid carcinoma: MR di-agnosis of lymph node metastasis. Am J Neuroradiol19:509–513.

18. Sackett DL, Haynes RB, Guyatt GH, Tugwell P 1991 The clin-ical examination. In: Sackett DL, Haynes RB, Guyatt GH,Tugwell P (eds) Clinical Epidemiology: A Basic Science for

Clinical Medicine. 2nd edition. Little, Brown and Company,Boston, pp 19–49.

19. Spiessl B, Beahrs OH, Hermanek P, Hutter RVP, Sceibe O,Sobin LH, Wagner G 1998 Head and neck tumours. In:Spiessl B, Beahrs OH, Hermanek P, Hutter RVP, Sceibe O,Sobin LH, Wagner G (eds) TNM Atlas: Illustrated Guide tothe TNM/pTNM classification of malignant tumours. 4thedition. Springer-Verlag, Tokyo, pp 3–61.

20. Cody HS III, Shah JP 1981 Locally invasive, well-differenti-ated thyroid cancer: 22 years’ experience at Memorial Sloan-Kettering Cancer Center. Am J Surg 142:480–483.

21. Fujimoto Y, Obara T, Ito Y, Kodama T, Yashiro T, YamashitaT, Nozaki M, Suzuki K 1986 Aggressive surgical approachfor locally invasive papillary carcinoma of the thyroid in pa-tients over forty-five years of age. Surgery 100:1098–1106.

22. Park CS, Suh KW, Min JS 1993 Cartilage-shaving procedurefor the control of tracheal cartilage invasion by thyroid car-cinoma. Head Neck 15:289–291.

Address reprint requests to:Shodayu Takashima, M.D.Department of Radiology

Shinshu University School of Medicine3-1-1 Asahi, Matsumoto 390-8621

Japan

E-mail: shodayu@hsp.md.shinshu-u.ac.jp

PROGNOSTIC SIGNIFICANCE OF MRI IN THYROID CANCER 1159