The radiologic work-up in thyroid surgery: Fine-needle biopsy versus scintigraphy and ultrasound.

Abstract

To compare the effectiveness and predictive value of radiologic studies with fine-needle aspiration biopsy (FNAB) in correctly diagnosing thyroid lesions, we reviewed the medical records of 441 patients who had been treated surgically for thyroid disease from 1987 through 1999. We compared the results of thyroid scintigraphy, ultrasound, and FNAB with findings on final surgical pathology. The data were analyzed according to the chi-squared ([chi square]) test. Of 189 thyroid scintigraphy scans that showed a hypofunctional (cold) nodule, 52 (27.5%) were found to be cancerous, and of 35 hyperfunctional (hot) nodules, two (5.7%) were malignant (sensitivity = 91%; specificity = 19%; accuracy = 38%; [chi square] = 7.67; p = 0.006). Of 66 ultrasounds that detected a solid or a mixed solid-cystic mass, 16(24.2%) were cancerous, while none of the eight sonograms that showed a purely cystic lesion was malignant (sensitivity = 100%; specificity = 14%; accuracy = 32%; [chi square] = 2.47; p = 0.116 [not statistically significant]). Of the 119 patients whose FNABs were diagnostic, 55 biopsies revealed follicular cells. Of the remaining 64 biopsies, cancer was correctly predicted in 35 of 44 patients (79.5%) and benign disease was correctly diagnosed in 18 of 20 patients (90.0%) (sensitivity = 95%; specificity = 67%; accuracy = 83%; [chi square] = 27.3; p = 0.00). We conclude that in the evaluation of thyroid lesions, FNAB is superior to imaging studies, which yield a relatively high rate of false-positive results.

Introduction

The most common indication for thyroid surgery is the excision of a mass or palpable nodule, which occurs in at least 4 to 7% of the population. (1) Even though most thyroid masses are benign, surgeons must still take steps to identify those patients whose lesions are potentially lethal. Many tests and evaluation protocols have been developed to help identify those patients who will benefit the most from thyroid surgery. Among them are scintigraphy, sonography, and fine-needle aspiration biopsy (FNAB). FNAB is inexpensive and diagnostically superior to the other modalities, (2) but its reliability depends on the experience and technical skill of the clinician who performs it. (3,4) For this reason, there continues to be a search for a better diagnostic procedure that yields more consistent results.

Okumura et al found thallium ([T1.sup.201]) planar scintigraphy easier to use and more accurate than FNAB. (5) Moreover, high-resolution ultrasound can resolve thyroid cysts and solid lesions as small as 2 mm. However, because its ability to correctly identify patients who do not have thyroid cancer is very limited, ultrasound is neither reliable nor cost-effective in the evaluation of patients with thyroid masses. (6) Even so, sonography is useful in guiding FNAB after a primary FNAB fails to yield a diagnosis, and it is helpful in evaluating patients whose thyroid masses are difficult to palpate. (7) Danese et al have actually proposed that ultrasound-guided FNAB be performed on all patients with thyroid masses at the beginning of the work-up to avoid the nondiagnostic results that can occur when FNAB is done without ultrasound assistance. (8)

In this article, we describe our study of the effectiveness and predictive value of radiologic studies and FNAB in the work-up of thyroid patients who are treated with surgery.

Patients and methods

We conducted a retrospective review of the medical records of all patients who underwent thyroid surgery from 1987 through 1999 at Hermann Hospital in Houston. We identified 441 such patients--65 males and 376 females, aged 13 to 91 years (mean: 42). Information was recorded on the clinical features, location, and pathology of the thyroid disease and on the type and method of work-up in preparation for thyroid surgery. Radiologic imaging and FNAB results were compared with the final pathology findings to determine the ability of each of these diagnostic methods to correctly predict the presence of cancer.

Because all variables were presented as frequencies, statistical analysis was performed in accordance with the chi-squared ([chi square]) test to determine if the actual and expected values were similar by chance. Statistical significance was reached when that probability was less than 5% (p<0.05).

Results

Preoperative diagnostic scintigraphy was performed on 243 patients, ultrasound on 83, and FNAB on 132 (some patients underwent more than one type of diagnostic test). Either thyroid enlargement or bilateral thyroid disease

was seen in 128 patients and unilateral disease in 310; among the latter group, the lesion was located in the left lobe in 144 patients and in the right lobe in 166 patients. In the remaining three patients, disease was located in the thyroid isthmus. The most common pathologies at the final diagnosis were goiter in 133 patients (30.2%), adenoma in 111(25.2%), and cancer in 100 (22.7%) (table 1). Among the 100 patients with a thyroid malignancy, 78 had papillary carcinoma, and they accounted for 17.7% of the entire study group.

Scintigraphy. Thyroid scintigraphy results were available for 243 patients. These scans identified a solitary hypofunctional (cold) nodule in 189 patients, a solitary hyperfunctional (hot) nodule in 35 patients, a goiter in nine patients, and normal findings in 10 (table 2).

Among the 189 patients with cold nodules, 52(27.5%) were found at final pathology to have cancer; of the 35 patients with hot nodules, cancer was confirmed in two (5.7%). In view of the fact that 137 of the 189 cold nodules were not cancerous, scintigraphic scanning yielded a false-positive rate of 72.5%; the false-negative rate for hot nodules was 5.7%. Overall, a cold nodule was associated with a significantly higher incidence of malignancy than was a hot nodule (sensitivity = 91%; specificity = 19%; accuracy = 38%; [chi square] = 7.67; p = 0.006).

We also compared the incidence of cancer among patients with cold nodules with its incidence among patients with all other findings on scintigraphy (table 2). Only five of the 54 patients (9.3%) in the latter group were found to have cancer at the final pathology. Therefore, a scan that identified a cold solitary nodule had a much greater chance of predicting a true malignancy than did a scan that demonstrated any other finding ([chi square] = 7.80; p = 0.005).

Ultrasound. Ultrasound results were available for 83 patients. A solid mass was identified on 56 of these images, a mixed solid-cystic mass on 10, a goiter on nine, and a pure cyst on eight (table 3).

Of the 56 solid masses, 15 (26.8%) were found to be malignant. Malignancies were also confirmed in one of the 10 mixed masses, in one of the nine goiters, and in none of the eight pure cysts. Even though solid masses on ultrasound were obviously more often associated with thyroid cancer than were purely cystic masses, the difference was not statistically significant ([chi square] = 2.80; p = 0.094).

When we combined the results of ultrasounds that indicated solid and mixed solid-cystic masses, we found that 16 of these 66 cases (24.2%) were associated with thyroid malignancy. Again, these masses were obviously more often associated with cancer than were the purely cystic masses but, again, the difference was not statistically significant (sensitivity = 100%, specificity = 14%, accuracy = 32%, [chi square] 2.47, p = 0.116).

Fine-needle aspiration biopsy. FNAB results were available for 132 patients. Of these, 119 were considered to be diagnostic and 13 were inconclusive (table 4). Twelve of these 13 were judged to be inconclusive even though sufficient aspirate had been obtained for analysis. Among the 12, final surgical pathology eventually revealed that nine patients had a follicular adenoma, two had nodular hyperplasia, and one had multinodular goiter. The one patient in whom sufficient aspirate was not available was later found to have papillary carcinoma. Our records did not show whether the 13 patients with nondiagnostic FNABs underwent a repeat biopsy.

Among the 119 patients in whom FNAB was considered to be diagnostic, follicular cells were seen in 55 patients, 10 of whom were confirmed as follicular carcinoma on final surgical pathology. FNAB indicated papillary carcinoma of the thyroid in 31 patients, 26 of whom (83.9%) did indeed have cancer. All three FNABs that suggested medullary carcinoma of the thyroid were correct.

When follicular neoplasms were excluded from analysis, the results of FNAB were better than those of any other test modality. Of the 20 masses that were deemed to be benign on FNAB, only two were eventually found to be malignant. Of the 44 nonfollicular masses that were predicted to be cancerous by FNAB, nine were found to be benign on final pathology (sensitivity = 95%; specificity = 67%; accuracy = 83%; [chi square] = 27.3; p = 0.00). The diagnosis of follicular carcinoma requires identification of capsular invasion, which is not possible on FNAB.

Discussion

Scintigraphy. Scintigraphy is useful in evaluating hyperthyroid patients who have a palpable thyroid nodule. A thyroid scan in these patients can determine whether the excess hormone is being produced in the nodule or in extranodular tissue. (9) The role of scintigraphy evolved to help determine which thyroid nodules or masses should undergo biopsy.

Other studies have shown that 6 to 10% of cold nodules are malignant, (10) but our study found that 52 of 189 cold nodules (27.5%) were cancerous. Additionally, two of 35 hot nodules (5.7%) on scintigraphy were malignant. Moreover, of the nine cases diagnosed as multinodular goiter by scintigraphy, only one (11.1%) was cancerous. When we combined all cases of noncold nodules and compared them with all cases of cold nodules, we found that only five of the 54 noncold nodules (9.3%) were malignant ([chi square] = 7.80; p = 0.005). Thyroid scintigraphy had a sensitivity of 91% in our patient population, but its specificity was only 19%. Thus, thyroid scintigraphy will not necessarily prevent unnecessary surgery in a patient with a thyroid mass.

In our study, hot nodules were found to be cancerous in two of 35 patients (5.7%). Therefore, scintigraphy is not useful in determining which thyroid nodules do not require biopsy because it can miss a malignancy in approximately 6% of hot nodules. The exception to this general rule pertains to autonomously functioning thyroid nodules, most of which are benign. (11)

Ultrasound. Sonographic examination of the thyroid can determine whether a thyroid mass is solid, cystic, or both. Sonography also provides more precise information--including an accurate determination of thyroid volume and structure--than does the best physical examination. High-resolution sonography can detect thyroid cysts as small as 2 mm. But overall, its ability to identify good surgical candidates without the assistance of other diagnostic modalities is very limited.

Generally, purely solid thyroid nodules are more likely to be malignant than are purely cystic masses. This was convincingly borne out in our study, as 15 of 56 solid nodules (26.8%) were found to be malignant, compared with none of the eight pure cysts. Results were similar when we calculated the combined malignancy rate of solid and mixed solid-cystic masses; 16 of these 66 masses (24.2%) were malignant. Another trend was evident with respect to goiter; only one of nine patients (11.1%) with sonographically identified goiter was found to have a malignant tumor.

Sonography did not miss a single case of malignancy in our study. The problem with it, however, is that it traditionally yields a high rate of false positives, and in our study, its specificity was only 14%. It is clear, then, that ultrasound alone is not sufficient to identify surgical candidates, and therefore it is not cost-effective.

Ultrasound might have a role in localizing small lesions that are not palpable on physical examination. But even then, the risk of cancer in small lesions is low, and even when one of these malignancies is present, it is usually not clinically significant. Finally, ultrasound can be used to guide a needle during aspiration biopsy of a nodule that is located far posteriorly, especially in an obese patient, but we did not encounter such a situation in our study.

Fine-needle aspiration biopsy. According to the literature, FNAB is the single most useful diagnostic tool for selecting patients for thyroid surgery, and it is now the procedure of choice in recently adopted diagnostic protocols. In our study, FNAB's true-positive rate was 79.5% (35/44) and its false-negative rate was 10.0% (2/20). When all malignant lesions were grouped together, FNAB had an overall sensitivity of 74%, a specificity of 88%, and an accuracy of 82%. Thus, compared with the other diagnostic modalities used in this study, FNAB was better at identifying patients who did not have cancer and who could therefore have been managed conservatively (provided they did not have any other indications for thyroid surgery). FNAB was even more reliable in patients with papillary cancer, which is the most common thyroid malignancy (sensitivity = 93%; specificity = 95%; accuracy = 94%; [chi square] 88.8; p = 0.00).

One of the drawbacks of FNAB is its relatively high incidence of inconclusive results and insufficient aspirate. In our study, this occurred in 13 of 132 patients (9.8%), a rate that is not higher than what has been reported in the literature. In nine of the 12 cases with inconclusive results, the final surgical diagnosis was follicular adenoma. This confirms the general consensus that follicular tumors cannot be reliably diagnosed by FNAB. A diagnosis of follicular carcinoma must be made by a complete inspection of the entire mass and evaluation of the tumor capsule. Two solutions to this problem would be the discovery of tumor markers and the development of new immunostaining techniques that could assist in the diagnosis of follicular tumors. Nevertheless, our data reaffirmed that FNAB is clearly the single most useful diagnostic test in the selection of patients for thyroid surgery, particularly in patients without follicular neoplasms.

Table 1.

Final surgical pathology in 441 patients who underwent thyroid surgery

Pathology           n (%)

Goiter            133 (30.2)
Adenoma           111 (25.2)
Cancer            100 (22.7)
 Papillary Ca      78
 Follicular Ca     10
 Medullary Ca       8
 Hurthle cell Ca    3
 Anaplastic Ca      1
Graves' disease    33 (7.5)
Thyroiditis        26 (5.9)
Colloid nodule     14 (3.2)
Other              24 (5.4)
Table 2.

Comparison of initial findings with final surgical pathology (n [%]) in
243 patients who underwent diagnostic scintigraphy

                          Final surgical pathology
Initial finding        Malignant             Benign

Cold nodule (n = 189)  52 (27.5)         137 (72.5)
Hot nodule (n = 35)      2 (5.7)          33 (94.3)
Goiter (n = 9)          1 (11.1)           8 (88.9)
Normal (n = 10)         2 (20.0)           8 (80.0)
Total (n = 243)        57 (23.5)         186 (76.5)
Table 3.

Comparison of initial findings with final surgical pathology (n [%]) in
83 patients who underwent diagnostic sonography

                                Final surgical pathology
Initial finding          Malignant                     Benign

Solid mass (n = 56)      15 (26.8)                   41 (73.2)
Mixed solid-cystic mass
(n = 10)                  1 (10.0)                    9 (90.0)
Goiter (n = 9)            1 (11.1)                    8 (88.9)
Pure cyst (n = 8)             0                      8 (100.0)
Total (n = 83)           17 (20.5)                   66 (79.5)
Table 4.

Comparison of initial findings with final surgical pathology (n [%]) in
132 patients who underwent diagnostic fine-needle aspiration biopsy

                                      Final surgical pathology
Initial finding                Malignant                    Benign

Follicular cells (n = 55)      10 (18.2)                   45 (81.8)
Papillary carcinoma (n = 31)   26 (83.9)                    5 (16.1)
Suspicious for cancer (n = 9)   5 (55.6)                    4 (44.4)
Medullary carcinoma (n = 3)    3 (100.0)                       0
Anaplastic carcinoma (n = 1)   1 (100.0)                       0
Benign lesion (n = 20)          2 (10.0)                   18 (90.0)
Inconclusive (n = 13)            1 (7.7)                   12 (92.3)
Total (n = 132)                48 (36.4)                   84 (63.6)

References

(1.) Vander JB, Gaston EA, Dawber TR. Significance of solitary nontoxic thyroid nodules: Preliminary report. N Engl J Med 1954;251:970-3.

(2.) Aschcraft MW, Van Herle AJ. Management of thyroid nodules. II: Scanning techniques, thyroid suppressive therapy, and fine needle aspiration. Head Neck Surg 1981;3:297-322.

(3.) Ross DS. Evaluation of the thyroid nodule. J Nucl Med 1991;32:2181-92.

(4.) Friedman M, Toriumi DM, Mafee MF. Diagnostic imaging techniques in thyroid cancer. Am J Surg 1988;155:215-23.

(5.) Okumura Y, Takeda Y, Sato S, et al. Comparison of differential diagnostic capabilities of 201T1 scintigraphy and fine-needle aspiration of thyroid nodules. J Nucl Med 1999;40:1971-7.

(6.) Dworkin HJ, Meier DA, Kaplan M. Advances in the management of patients with thyroid disease. Semin Nucl Med 1995;25:205-20.

(7.) Sanchez RB, vanSonnenberg E, D'Agostino HB, et al. Ultrasound guided biopsy of nonpalpable and difficult to palpate thyroid masses. J Am Coll Surg 1994;178:33-7.

(8.) Danese D, Sciacchitano S, Farsetti A, et al. Diagnostic accuracy of conventional versus sonography-guided fine-needle aspiration biopsy of thyroid nodules. Thyroid 1998;8:15-21.

(9.) Capri A. Is thyroid scintigraphy still useful? Endocr Surg 1990;7:399-403.

(10.) Freitas JE, Freitas AE. Thyroid and parathyroid imaging. Semin Nucl Med 1994;24:234-45.

(11.) Meier DA, Dworkin HJ. The autonomously functioning thyroid nodule. J Nucl Med 1991;32:30-2.

From the Department of Otolaryngology-Head and Neck Surgery, University of Virginia Medical School, Charlottesville (Dr. Kountakis), Venizelion Hospital, Crete, Greece (Dr. Skoulas), and the University of Texas-Houston Medical School (Dr. Maillard).

Reprint requests: Stilianos E. Kountakis, MD, Associate Professor, Department of Otolaryngology-Head and Neck Surgery, University of Virginia Medical School, P.O. Box 800713, Charlottesville, VA 22908. Phone: [phone]; fax: [phone]; e-mail: [e-mail]

Originally presented at a plenary session at the annual meeting of the American Academy of Otolaryngology-Head and Neck Surgery; Washington, D.C.; Sept. 27, 2000.