Role of Ultrasound in the Management of Thyroid Nodules David A. K. Watters, FRCSEd, Anil T. Ahuja, MD, Rhodri M. Evans, ~CR, Wilson Chick, MRCPath, Walter W. K. King, MD, Constantine Mctrewcli, FRCR,Arthur K. C. Li, MD, HongKong
One hundred twenty patients undergoing thyroid surgery for thyroid nodules or goiter were examined by preoperative ultrasound and fine needle aspiration (FNA) cytology. In the determination of whether a lesion was malignant, FNA had sensitivity, specificity, and positive predictive values of 86%, 85%, and 58%, respeetively. Ultrasound had sensitivity, speeffieity, and positive predictive values of 74%, 83%, and 51%, respectively. The different types of thyroid pathology showed different ultrasonic features in most cases, although no single feature was pathognomonie. Malignant lesions tended to be solid and hypoechoic without a halo, but there was a cystic element in 26% of the lesions and calcification in 37%. Ultrasound was superior to FNA in diagnosing nodular goiter with sensitivity, specificity, and positive predietive values of 70%, 93%, and 92%, respectively, compared with 55%, 86%, and 83%, respectively. The two modalities are complementary.
he most common reason for operating on a patient with a thyroid nodule is the risk of malignancy, T which will be present in only 10% to 20% of patients
[1-3]. Clinical examination alone often fails to differentiate between benign and malignant lesions [3]. Fine necdie aspiration (FNA) of a nodule enables the correct preoperative diagnosis to be made in at least 75% of the cases and in over 90% of the nodules that arc malignant [4]. Follow-up of patients with benign pathology suggests FNA has a falsc-negative rate of 0.7% for malignancy [5]. However, FNA fails to provide a satisfactory specimen in about 15% of cases and may be particularly difficult to perform if the nodule is small, and, in any case, FNA samples only one area of the thyroid. Although ultrasound allows the whole thyroid and adjacent lymph nodes to be examined, it has been regarded as valuable only in differentiatingbetween solid and cystic lesions, and single and multiple lesions,but not in identifying different pathologies [6-7]. This study was undertaken to assess the accuracy of ultrasound in the diagnosis of thyroid nodule pathology and to compare the diagnostic accuracy of ultrasound with FNA.
PATIENTS AND METHODS The ultrasound and case records of 120 patients who underwent thyroid surgery between January 1989 and August 1991 were reviewed. Patients were originally sccn in the thyroid clinicwhere they were assessed clinically, and thyroid function testsand an F N A were performed. Smears were fixedusing Cytof'Lx(Salmond Smith Biolab Ltd., Auckland, N e w Zealand), and the restof the aspirate was sent in 50% alcohol for cellblock. Ultrasound examination was requested, and sometimes F N A would be repeated under ultrasound guidance (46 cases).F N A was repeated at follow-up when insufficientmaterial for diagnosis was obtained. The ultrasound examinations were supervised and reviewed without any knowledge of the finalor F N A diagnosisby two of the authors (ATA, RME). Staticscans (68) were performed using a Phillips (The Netherlands) 7,300 static sonodiaguostic system and a 10-MHz transducer. Real-time scans (52) were performed by the same two authors using 7.5-MHz linear or 10-MHz sector probes with an Aloka 650 scanner. The features of the thyroid nodule or nodules were Fromthe Departmentsof Surgery(DAKW,WWKK,AKCL), Radiol- described in terms of echogcnicity (reduced, normal, or ogy (ATA, RME, CM), and Pathology(WC), Prince of Wales Hospi- increased), gcncity (homo or hetero), cystic component tal, ChineseUniversityof Hong Kong,Hong Kong. (nil,lessthan 50%, equal to 50%, or greater than 50%), Requests for reprintsshouldbe addressed to Arthur K. C. Li, MD, calcification(present or absent), halo appearance (preDepartment of Surgery, Prince of Wales Hospital, ChineseUniversity sent, absent, or doubtful), and noduladty (solitaryor of Hong Kong,Shatin, NT, Hong Kong. Presentedat the Third InternationalConferenceon Head and Neck multiple). Cancer, San Francisco,California,April 26-30, 1992. The clinicalpresentation, pathologic diagnosis, and 654
THE AMERICAN JOURNAL OF SURGERY
VOLUME 164 DECEMBER 1992
ULTRASOUND IN THYROID NODULES
TABLE 1
T A B L E II
Sensitivity, Specificity, and Predictive Value of Fine
Sensitivity, Specificity, a n d P r e d i c t i v e V a l u e of
Needle Aspiration for Thyroid Pathology
Ultrasound for Thyroid Pathology
Diagnosis Malignancy Follicular adenoma Nodular goiter Cyst Total
NegaPositive tive Predic- PredicNo. of Sensi- Specitive tive AccuAspir- tivity ficity Value Value racy ations (%) (%) (%) (%) (%) 23 28
86 72
64 5 120
55 . .
85 81
58 54
86 . .
83 .
.
.
96 90
84 71
61
69
Diagnosis Malignancy Follicular adenoma Nodular goiter Cyst Total
.
.
69
NegaPositive tive Predic- PredicNo. of Sensi- Specitive tive AccuUltrativity ficity Value Value racy sounds (%) (%) (%) (%) (%) 23 28
74 75
64 5 120
70 . .
83 83
51 57
93 . .
92 .
.
.
93 91
82 81
73
81
.
.
81
TABLE Ill
Ultrasound Features (%)
Diagnosis Papillary carcinoma Follicular neoplasm* Multinodular goiter
Halo Calcification Ab- IncomAbEchogenicity Geneity Cystic Component Present sent plete Present sent Hyper Iso Hypo Homo Hetero None 50% 10 52 35
79 36 60
11 12 6
37 4 25
63 96 75
0 32 13
10 32 33
89 36 54
42 40 8
58 60 92
74 52 44
21 32 35
5 4 10
0 12 11
Nodularity Soil- Multary tiple 83 89 44
17 11 56
*Includes follicularadenomaand follicularcarcinoma.
outcome of treatment were reviewed using the case notes and follow-up in the thyroid clinic. The term nodular goiter has been used to include the pathologic diagnosis of nodular hyperplasia, adenomatous goiter, and colloid nodule. RESULTS The nodules of the 120 patients were divided into 4 diagnostic groups: malignant (19 papillary, 3 follicular carcinoma, 1 medullary), follicular adenoma (28), nodular goiter (64), and cystic (5). The patients were between 13 to 84 years of age. The nodules ranged in size between 1 to 11 cm and had a duration of 1 month to 40 years. FNA: FNA was not performed in six patients. In nine patients (7.5%), an insufficient number of cells for diagnosis were obtained, even after repeated aspiration (two papiliary carcinomas, two follicular adenomas, and five nodular goiters). The correct diagnosis was made by FNA in 73 of the remaining 105 patients (69%). FNA was most sensitive (86%) and specific (85%) in diagnosing malignant lesions but less sensitive in diagnosing nodular goiter or follicular adenoma (Table I). A FNA diagnosis of follicular neoplasm or proliferative follicular lesion was made in 12 patients with nodular goiter, 12 patients with follicular adenoma, and 4 patients with malignancy (2 papillary, 1 follicular variant of papillary, and 1 follicular carcinoma). FNA was performed in four of five patients
with cysts. The correct diagnosis was made in three patients, but in one there was a false-positive result that indicated the presence of a malignancy. Ultrasound: Ultrasound examination was performed in 120 patients. The correct pathology was diagnosed in 97 patients (81%). Ultrasound was less sensitive than FNA in diagnosing malignancy (74% versus 86%) but more sensitive in diagnosing follicular adenoma (75% versus 72%) and nodular goiter (70% versus 55%) (Table II). Ultrasound diagnosed the cystic component in all five cysts, hut there also appeared to be a solid component that was not conf'Lrmed by histologic examination. Malignant lesions tend to be solid and hypoechoic without a halo (Table HI). Calcification was present in 37% of malignancies, and, in 26% of malignancies, there was a cystic component of up to 50%. Follicular aden~ mas (Figure 1) rarely contained calcium, were of mixed echogenicity, but were cystic and possessed a halo in about 50% of cases. Nodular goiters had multiple nodules in 34 cases (56%), calcified in a fourth of cases, and possessed a halo in a third. In 3 of 23 malignancies, both FNA and ultrasound diagnoses were incorrect. Ultrasound correctly diagnosed malignancy in two patients in whom FNA failed to achieve the correct diagnosis (one insufficient aspirate), and FNA diagnosed three patients with malignancy in whom the ultrasound indicated a benign lesion.
THE AMERICAN JOURNAL OF SURGERY
VOLUME 164 DECEMBER 1992
655
W A I T E R S ET AL
Figure 1. A follicular adenoma (as shown in a longitudinal section by ultrasound) is found to be an isoechoic, homogeneous, welldefined, noncalclfied solid mass with a halo (arrow).
COMMENTS This study shows that, despite recent reports to the contrary, different thyroid pathologies do have specific ultrasonic features that allow a correct diagnosis to be made in 81% of cases with sensitivity, specificity and positive predictive values for malignancy of 74%, 83%, and 51%, respectively. Jones et al [8] found that the sensitivity, specificity, and positive predictive values of ultrasound in suggesting a malignant diagnosis were 75%, 61%, and 19%, respectively. However, they interpreted an ultrasound report as suggestive of malignancy if the nodule was solid or of a mixed solid/cystic structure. We maintain that a malignant lesion should be suspected when there is a hypoechoic, nonhaloed lesion that may have a cystic component of 50% or less. Calcification may be present in a third of lesions. Another advantage of ultrasound is that it allows the whole gland to be examined, rather than just the dominant nodule. However, ultrasound is limited by the fact that no features are pathognomonic for malignancy. It is thus complementary, rather than an alternative, to FNA for thyroid nodules. FNA was slightly better at diagnosing malignancy and in making a specific pathologic diagnosis, particularly when Hfirthle ceils or papillary cells were present. Ultrasound also enables an accurate aspiration to be done under direct vision. When a thyroid nodule is small and virtually impalpable, ultrasound-guided aspiration is the best way of obtaining cells for cytology. A recent patient in our ward presented with an enlarged cervical lymph node that was shown to be a papillary carcinoma by FNA. The thyroid primary carcinoma was detected on ultrasound and confirmed by ultrasound-guided FNA. FNA may sometimes result in a thyroid hematoma, which will make specific ultrasonographic features within the original nodule difficult to recognize. If thyroid ultra-
656
sound can be performed without delay, it might be best to perform the first FNA under ultrasound guidance. However, in many hospitals, this would not be practical because of the demands on the radiology services. When ultrasound and FNA concur, one can be confident that the diagnosis is correct. If FNA diagnoses malignancy but ultrasound does not, we advise surgery. The final choice of procedure will then be made after frozen section analysis of a specimen. The least accurate, potentially malignant FNA diagnoses are follicular neoplasm and proliferative follicular lesion, and, with these diagnoses, if ultrasound shows a multinodular goiter, it is probably safe to repeat the FNA in the expectation of obtaining a diagnosis of nodular hyperplasia or adenomatous goiter. Whenever there is doubt about the possibility of malignancy, it is safer to operate. Clinical signs that suggest malignancy include a solid, hard nodule, especially if it is larger than 2 cm, or there is a palpable lymph node or concurrent recurrent laryngeal nerve palsy. Patients with these signs warrant surgery, regardless of the FNA or ultrasound results. However, the majority of patients do not have such clinical signs, and, in these patients, a firm diagnosis should be possible with FNA and ultrasound in approximately 80% of cases so that most patients with benign disease can be observed. We are uncertain how long the patients should be followed but the Mayo Clinic experience showed that only 3 of 439 patients with a benign FNA turned out to have or develop malignancy during a 6-year follow-up [5]. Thus, the possibility of missing a malignancy in a patient in whom ultrasound and FNA concur is small, and, unless the patient is anxious to have the nodule removed, it is probably safe to observe those with nodules that are static or decreasing in size.
THE AMERICANJOURNALOF SURGERY VOLUME164 DECEMBER1992
ULTRASOUND IN THYROID NODULES
Better definition of pathologic features should develop with more experience and feedback between radiologists, pathologists, and surgeons. Further prospective studies are indicated to evaluate the precise pathology of some ultrasound features, such as the halo, and also to follow patients with benign features who do not undergo surgery.
REFERENCES 1. Brooks JR, Starnes F, Brooks DC, Pelkey JN. Surgical therapy for thyroid carcinoma: a review of 1249 solitary thyroid nodules. Surgery 1988; 104: 940-6. 2. Christensen SB, Bondenson L, Ericsson UB, Lindholm K. Prediction of malignancy in the solitary thyroid nodule by physical examination, thyroid scan, fine needle biopsy and serum thyrogiobulin. A prospective study of 100 surgically treated patients. Aeta
Chit Scand 1984; 150: 433-9. 3. Brager R J, Silver CE. Needle aspiration biopsy of thyroid nodules. Laryngoscope 1984; 94: 38-42. 4,. Boey J, Hsu C, Wong J, Ong GB. Fine needle aspiration venus drill needle biopsy of thyroid nodules: a controlled clinical trial. Surgery 1982; 91: 611-5. 5. Grant CS, Hay ID, Gough IR, McCarthy PM, Gcollner JR. Long-term follow-up of patients with benign thyroid fme needle aspiration cytologic diagnoses. Surgery 1989; 106: 980-6. 6. Cox MR, Marshall SG, Spence RAJ. Solitary thyroid nodule: a prospective evaluation of nuclear scanning and ultrasonography. Br J Surg 1991; 78: 90-3. 7. A1Sayer HM, Bayliss AP, Krukowski ZH, Matheson NA. The limitations of ultrasound in thyroid swellings. JR (2oll Surg Edinb 1986; 31: 27-31. 8. Jones AJ, Airman T J, Edmonds CJ, Burke M, Hudson E, Tellez M. Comparison of fine needle aspiration cytology, radioisotopic and ultrasound scanning in the management of thyroid nodules. Postgrad Med J 1990; 66: 914-7.
THE AMERICAN JOURNAL OF SURGERY VOLUME 164 DECEMBER 1992 657
One hundred twenty patients undergoing thyroid surgery for thyroid nodules or goiter were examined by preoperative ultrasound and fine needle aspiration (FNA) cytology. In the determination of whether a lesion was malignant, FNA had sensitivity, specificity, and positive predictive values of 86%, 85%, and 58%, respeetively. Ultrasound had sensitivity, speeffieity, and positive predictive values of 74%, 83%, and 51%, respectively. The different types of thyroid pathology showed different ultrasonic features in most cases, although no single feature was pathognomonie. Malignant lesions tended to be solid and hypoechoic without a halo, but there was a cystic element in 26% of the lesions and calcification in 37%. Ultrasound was superior to FNA in diagnosing nodular goiter with sensitivity, specificity, and positive predietive values of 70%, 93%, and 92%, respectively, compared with 55%, 86%, and 83%, respectively. The two modalities are complementary.
he most common reason for operating on a patient with a thyroid nodule is the risk of malignancy, T which will be present in only 10% to 20% of patients
[1-3]. Clinical examination alone often fails to differentiate between benign and malignant lesions [3]. Fine necdie aspiration (FNA) of a nodule enables the correct preoperative diagnosis to be made in at least 75% of the cases and in over 90% of the nodules that arc malignant [4]. Follow-up of patients with benign pathology suggests FNA has a falsc-negative rate of 0.7% for malignancy [5]. However, FNA fails to provide a satisfactory specimen in about 15% of cases and may be particularly difficult to perform if the nodule is small, and, in any case, FNA samples only one area of the thyroid. Although ultrasound allows the whole thyroid and adjacent lymph nodes to be examined, it has been regarded as valuable only in differentiatingbetween solid and cystic lesions, and single and multiple lesions,but not in identifying different pathologies [6-7]. This study was undertaken to assess the accuracy of ultrasound in the diagnosis of thyroid nodule pathology and to compare the diagnostic accuracy of ultrasound with FNA.
PATIENTS AND METHODS The ultrasound and case records of 120 patients who underwent thyroid surgery between January 1989 and August 1991 were reviewed. Patients were originally sccn in the thyroid clinicwhere they were assessed clinically, and thyroid function testsand an F N A were performed. Smears were fixedusing Cytof'Lx(Salmond Smith Biolab Ltd., Auckland, N e w Zealand), and the restof the aspirate was sent in 50% alcohol for cellblock. Ultrasound examination was requested, and sometimes F N A would be repeated under ultrasound guidance (46 cases).F N A was repeated at follow-up when insufficientmaterial for diagnosis was obtained. The ultrasound examinations were supervised and reviewed without any knowledge of the finalor F N A diagnosisby two of the authors (ATA, RME). Staticscans (68) were performed using a Phillips (The Netherlands) 7,300 static sonodiaguostic system and a 10-MHz transducer. Real-time scans (52) were performed by the same two authors using 7.5-MHz linear or 10-MHz sector probes with an Aloka 650 scanner. The features of the thyroid nodule or nodules were Fromthe Departmentsof Surgery(DAKW,WWKK,AKCL), Radiol- described in terms of echogcnicity (reduced, normal, or ogy (ATA, RME, CM), and Pathology(WC), Prince of Wales Hospi- increased), gcncity (homo or hetero), cystic component tal, ChineseUniversityof Hong Kong,Hong Kong. (nil,lessthan 50%, equal to 50%, or greater than 50%), Requests for reprintsshouldbe addressed to Arthur K. C. Li, MD, calcification(present or absent), halo appearance (preDepartment of Surgery, Prince of Wales Hospital, ChineseUniversity sent, absent, or doubtful), and noduladty (solitaryor of Hong Kong,Shatin, NT, Hong Kong. Presentedat the Third InternationalConferenceon Head and Neck multiple). Cancer, San Francisco,California,April 26-30, 1992. The clinicalpresentation, pathologic diagnosis, and 654
THE AMERICAN JOURNAL OF SURGERY
VOLUME 164 DECEMBER 1992
ULTRASOUND IN THYROID NODULES
TABLE 1
T A B L E II
Sensitivity, Specificity, and Predictive Value of Fine
Sensitivity, Specificity, a n d P r e d i c t i v e V a l u e of
Needle Aspiration for Thyroid Pathology
Ultrasound for Thyroid Pathology
Diagnosis Malignancy Follicular adenoma Nodular goiter Cyst Total
NegaPositive tive Predic- PredicNo. of Sensi- Specitive tive AccuAspir- tivity ficity Value Value racy ations (%) (%) (%) (%) (%) 23 28
86 72
64 5 120
55 . .
85 81
58 54
86 . .
83 .
.
.
96 90
84 71
61
69
Diagnosis Malignancy Follicular adenoma Nodular goiter Cyst Total
.
.
69
NegaPositive tive Predic- PredicNo. of Sensi- Specitive tive AccuUltrativity ficity Value Value racy sounds (%) (%) (%) (%) (%) 23 28
74 75
64 5 120
70 . .
83 83
51 57
93 . .
92 .
.
.
93 91
82 81
73
81
.
.
81
TABLE Ill
Ultrasound Features (%)
Diagnosis Papillary carcinoma Follicular neoplasm* Multinodular goiter
Halo Calcification Ab- IncomAbEchogenicity Geneity Cystic Component Present sent plete Present sent Hyper Iso Hypo Homo Hetero None 50% 10 52 35
79 36 60
11 12 6
37 4 25
63 96 75
0 32 13
10 32 33
89 36 54
42 40 8
58 60 92
74 52 44
21 32 35
5 4 10
0 12 11
Nodularity Soil- Multary tiple 83 89 44
17 11 56
*Includes follicularadenomaand follicularcarcinoma.
outcome of treatment were reviewed using the case notes and follow-up in the thyroid clinic. The term nodular goiter has been used to include the pathologic diagnosis of nodular hyperplasia, adenomatous goiter, and colloid nodule. RESULTS The nodules of the 120 patients were divided into 4 diagnostic groups: malignant (19 papillary, 3 follicular carcinoma, 1 medullary), follicular adenoma (28), nodular goiter (64), and cystic (5). The patients were between 13 to 84 years of age. The nodules ranged in size between 1 to 11 cm and had a duration of 1 month to 40 years. FNA: FNA was not performed in six patients. In nine patients (7.5%), an insufficient number of cells for diagnosis were obtained, even after repeated aspiration (two papiliary carcinomas, two follicular adenomas, and five nodular goiters). The correct diagnosis was made by FNA in 73 of the remaining 105 patients (69%). FNA was most sensitive (86%) and specific (85%) in diagnosing malignant lesions but less sensitive in diagnosing nodular goiter or follicular adenoma (Table I). A FNA diagnosis of follicular neoplasm or proliferative follicular lesion was made in 12 patients with nodular goiter, 12 patients with follicular adenoma, and 4 patients with malignancy (2 papillary, 1 follicular variant of papillary, and 1 follicular carcinoma). FNA was performed in four of five patients
with cysts. The correct diagnosis was made in three patients, but in one there was a false-positive result that indicated the presence of a malignancy. Ultrasound: Ultrasound examination was performed in 120 patients. The correct pathology was diagnosed in 97 patients (81%). Ultrasound was less sensitive than FNA in diagnosing malignancy (74% versus 86%) but more sensitive in diagnosing follicular adenoma (75% versus 72%) and nodular goiter (70% versus 55%) (Table II). Ultrasound diagnosed the cystic component in all five cysts, hut there also appeared to be a solid component that was not conf'Lrmed by histologic examination. Malignant lesions tend to be solid and hypoechoic without a halo (Table HI). Calcification was present in 37% of malignancies, and, in 26% of malignancies, there was a cystic component of up to 50%. Follicular aden~ mas (Figure 1) rarely contained calcium, were of mixed echogenicity, but were cystic and possessed a halo in about 50% of cases. Nodular goiters had multiple nodules in 34 cases (56%), calcified in a fourth of cases, and possessed a halo in a third. In 3 of 23 malignancies, both FNA and ultrasound diagnoses were incorrect. Ultrasound correctly diagnosed malignancy in two patients in whom FNA failed to achieve the correct diagnosis (one insufficient aspirate), and FNA diagnosed three patients with malignancy in whom the ultrasound indicated a benign lesion.
THE AMERICAN JOURNAL OF SURGERY
VOLUME 164 DECEMBER 1992
655
W A I T E R S ET AL
Figure 1. A follicular adenoma (as shown in a longitudinal section by ultrasound) is found to be an isoechoic, homogeneous, welldefined, noncalclfied solid mass with a halo (arrow).
COMMENTS This study shows that, despite recent reports to the contrary, different thyroid pathologies do have specific ultrasonic features that allow a correct diagnosis to be made in 81% of cases with sensitivity, specificity and positive predictive values for malignancy of 74%, 83%, and 51%, respectively. Jones et al [8] found that the sensitivity, specificity, and positive predictive values of ultrasound in suggesting a malignant diagnosis were 75%, 61%, and 19%, respectively. However, they interpreted an ultrasound report as suggestive of malignancy if the nodule was solid or of a mixed solid/cystic structure. We maintain that a malignant lesion should be suspected when there is a hypoechoic, nonhaloed lesion that may have a cystic component of 50% or less. Calcification may be present in a third of lesions. Another advantage of ultrasound is that it allows the whole gland to be examined, rather than just the dominant nodule. However, ultrasound is limited by the fact that no features are pathognomonic for malignancy. It is thus complementary, rather than an alternative, to FNA for thyroid nodules. FNA was slightly better at diagnosing malignancy and in making a specific pathologic diagnosis, particularly when Hfirthle ceils or papillary cells were present. Ultrasound also enables an accurate aspiration to be done under direct vision. When a thyroid nodule is small and virtually impalpable, ultrasound-guided aspiration is the best way of obtaining cells for cytology. A recent patient in our ward presented with an enlarged cervical lymph node that was shown to be a papillary carcinoma by FNA. The thyroid primary carcinoma was detected on ultrasound and confirmed by ultrasound-guided FNA. FNA may sometimes result in a thyroid hematoma, which will make specific ultrasonographic features within the original nodule difficult to recognize. If thyroid ultra-
656
sound can be performed without delay, it might be best to perform the first FNA under ultrasound guidance. However, in many hospitals, this would not be practical because of the demands on the radiology services. When ultrasound and FNA concur, one can be confident that the diagnosis is correct. If FNA diagnoses malignancy but ultrasound does not, we advise surgery. The final choice of procedure will then be made after frozen section analysis of a specimen. The least accurate, potentially malignant FNA diagnoses are follicular neoplasm and proliferative follicular lesion, and, with these diagnoses, if ultrasound shows a multinodular goiter, it is probably safe to repeat the FNA in the expectation of obtaining a diagnosis of nodular hyperplasia or adenomatous goiter. Whenever there is doubt about the possibility of malignancy, it is safer to operate. Clinical signs that suggest malignancy include a solid, hard nodule, especially if it is larger than 2 cm, or there is a palpable lymph node or concurrent recurrent laryngeal nerve palsy. Patients with these signs warrant surgery, regardless of the FNA or ultrasound results. However, the majority of patients do not have such clinical signs, and, in these patients, a firm diagnosis should be possible with FNA and ultrasound in approximately 80% of cases so that most patients with benign disease can be observed. We are uncertain how long the patients should be followed but the Mayo Clinic experience showed that only 3 of 439 patients with a benign FNA turned out to have or develop malignancy during a 6-year follow-up [5]. Thus, the possibility of missing a malignancy in a patient in whom ultrasound and FNA concur is small, and, unless the patient is anxious to have the nodule removed, it is probably safe to observe those with nodules that are static or decreasing in size.
THE AMERICANJOURNALOF SURGERY VOLUME164 DECEMBER1992
ULTRASOUND IN THYROID NODULES
Better definition of pathologic features should develop with more experience and feedback between radiologists, pathologists, and surgeons. Further prospective studies are indicated to evaluate the precise pathology of some ultrasound features, such as the halo, and also to follow patients with benign features who do not undergo surgery.
REFERENCES 1. Brooks JR, Starnes F, Brooks DC, Pelkey JN. Surgical therapy for thyroid carcinoma: a review of 1249 solitary thyroid nodules. Surgery 1988; 104: 940-6. 2. Christensen SB, Bondenson L, Ericsson UB, Lindholm K. Prediction of malignancy in the solitary thyroid nodule by physical examination, thyroid scan, fine needle biopsy and serum thyrogiobulin. A prospective study of 100 surgically treated patients. Aeta
Chit Scand 1984; 150: 433-9. 3. Brager R J, Silver CE. Needle aspiration biopsy of thyroid nodules. Laryngoscope 1984; 94: 38-42. 4,. Boey J, Hsu C, Wong J, Ong GB. Fine needle aspiration venus drill needle biopsy of thyroid nodules: a controlled clinical trial. Surgery 1982; 91: 611-5. 5. Grant CS, Hay ID, Gough IR, McCarthy PM, Gcollner JR. Long-term follow-up of patients with benign thyroid fme needle aspiration cytologic diagnoses. Surgery 1989; 106: 980-6. 6. Cox MR, Marshall SG, Spence RAJ. Solitary thyroid nodule: a prospective evaluation of nuclear scanning and ultrasonography. Br J Surg 1991; 78: 90-3. 7. A1Sayer HM, Bayliss AP, Krukowski ZH, Matheson NA. The limitations of ultrasound in thyroid swellings. JR (2oll Surg Edinb 1986; 31: 27-31. 8. Jones AJ, Airman T J, Edmonds CJ, Burke M, Hudson E, Tellez M. Comparison of fine needle aspiration cytology, radioisotopic and ultrasound scanning in the management of thyroid nodules. Postgrad Med J 1990; 66: 914-7.
THE AMERICAN JOURNAL OF SURGERY VOLUME 164 DECEMBER 1992 657
