| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Departments of Surgery (ADB, ARS, WD, MFB, JS) and Pathology (AGH, MZ), Memorial Sloan-Kettering Cancer Center, New York, New York.
Correspondence: Address correspondence and reprint requests to: Dr. A. R. Shaha, 1275 York Ave., New York, NY 10021; Fax: 212-717-3302; E-mail: shahaa{at}mskcc.org
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Methods: All patients who had operations for thyroid disease between January 1996 and June 1999 were identified in our prospective database. Completion and incidental thyroidectomies were excluded. Data obtained from the pathology files included FNA, FS, and the final histologic diagnosis.
Results: Five hundred sixty-four patients, including 409 women (73%), with a median age of 50 years (range, 6–94) were identified, of whom 293 (52%) had cancer diagnosed on permanent sections. Three hundred twenty-nine patients (58%) had evaluable FNA, of which 91 (28%) were benign, 94 were malignant (28%), and 144 (44%) were suspicious (46% of these were malignant on final). Frozen section was performed in 397 (70%) patients; of these samples, 170 (43%) were found to be benign, 106 (27%) were malignant, and 121 (30%) were deferred (46% malignant on final). Fine-needle aspiration positively identified 51% of confirmed malignancies; 13% of patients with malignancy had a benign FNA result. Total thyroidectomy was performed in 64% of malignant tumors and 29% of benign thyroid disease (P < .001). Logistic regression revealed no association of extent of surgery with FNA results. A frozen section positive for malignancy was associated with total thyroidectomy (P < .001, RR 6 [CI 3–10]), and a negative frozen section report was associated with lobectomy (P < .05, RR 0.5 [CI 0.3–0.96]). Frozen sections results altered the preoperative plan in only 29 patients (5%).
Conclusion: Results of preoperative FNA had no direct impact on the selection of the surgical procedure in this selected cohort. Intraoperative FS added very little to surgical management. The majority of thyroid operations at this institution are planned and performed based on known prognostic factors and intraoperative findings.
Key Words: Thyroid neoplasms • Thyroid surgery • Needle biopsy • Frozen section
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
There is a large body of literature discussing the relative merits of fine-needle aspiration biopsy (FNA) for the preoperative identification of malignancy.3–13 Although FNA is a cost-effective and specific first test, up to 30% of patients with nodular thyroid disease will receive an FNA report that reads "follicular lesion" or "suspicious." Most of these patients will require surgical management, and the extent of resection will need to be planned without a preoperative diagnosis of malignancy.
Whereas an overlapping body of literature discusses the use of intraoperative frozen section for the confirmation or clarification of preoperative FNA results,7,9,12,14–17 whereas other authors have questioned the benefit of the intraoperative frozen section over FNA.3,8,10,18–20 At our institution, a frozen section may be reported as "follicular lesion" or "defer to permanent (defer reporting results until permanent sections are reviewed and reported)" in up to one third of patients. In these patients the extent of resection must be decided without intraoperative confirmation of malignancy.
Several groups have described prognostic features that aid in planning the extent of thyroid resection.21–23 In reviewing our experience at Memorial Sloan-Kettering Cancer Center, we previously have demonstrated that factors such as age, gender, history of radiation exposure, tumor size, and presence of metastatic disease in combination with grade and histologic type of malignancy are strong determinants of survival.24–26 We place our patients in a low-, intermediate-, or high-risk group based on these criteria. In high-risk patients, where subsequent therapy will require radioactive iodine, we perform a total thyroidectomy. In low-risk patients, a lobectomy with isthmectomy is adequate therapy.27 For intermediate-risk patients, we use an individualized approach.24 In addition, the intraoperative examination of the thyroid gland and surrounding soft tissues adds more information for operative planning. Patients with bilateral nodules or findings suspicious for extracapsular spread are safely treated by total thyroidectomy.
The purpose of this study was to determine the value of fine-needle aspiration and frozen section analysis for patients who are selected for thyroidectomy based on risk stratification, and to examine their importance in determining the extent of thyroid resection.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Permanent pathologic diagnoses were coded as malignant for any report of papillary, follicular, Hürthle cell, medullary, or anaplastic thyroid carcinoma. Microcarcinomas were included with papillary cancers. Lymphoma and metastatic disease involving the thyroid gland were coded as malignant. The maximum diameter of the tumor or the largest of multiple tumors was recorded for all malignant lesions. All remaining diagnoses were recorded as benign, including goiter, adenoma, Hürthle cell adenoma, hyperplasia, and thyroiditis.
Fine-needle aspiration results were considered malignant for a written diagnosis of thyroid carcinoma, lymphoma, or metastatic disease. FNA was recorded as "none" if there was no result listed in the computer. This category included patients who had outside FNA biopsies that were never submitted to Memorial Sloan-Kettering Cancer Center for cytologic review. An unsatisfactory specimen was coded as such and included with the "none" category for most analyses. A report listed as "benign," "cyst," "goiter," "nodule," "adenoma," or "no evidence of malignancy" was recorded as benign. Diagnoses of "Hürthle cell," "possible" or "suspicious" for any malignancy, "neoplasm," or "lesion" were recorded as suspicious.
Frozen section diagnoses of the thyroid gland were included in the analysis. Results were coded as malignant for all written reports of thyroid cancer, lymphoma, or metastatic disease. Reports including "favor papillary cancer," "possible papillary cancer," or "suspicious for capsular invasion" were coded as malignant. Goiter, hyperplasia, thyroiditis, or adenomas were considered benign. In addition, diagnoses beginning with "benign" or describing a "nodule" were coded as benign. All deferred diagnoses except "defer hyperplasia" and "defer thyroiditis" were coded as deferred. In addition, all diagnoses of "tumor," "follicular lesion," "follicular neoplasm," "Hürthle lesion," or "Hürthle neoplasm" were considered deferred.
Results of FNA biopsy and frozen section were compared to permanent histologic examination for determinations of sensitivity, specificity, and predictive value. First, all suspicious or deferred results were grouped with the positive results, and sensitivity and specificity were calculated (calculation A). Second, the suspicious or deferred results were grouped with the negative results, and sensitivity and specificity were recalculated (calculation B). Both pairs of sensitivity and specificity are reported for each test. Positive predictive value was based on the equation: (true positives) ÷ (true and false positives); negative predictive value was calculated by the equation: (true negatives) ÷ (true and false negatives). These results are reported for calculations A and B. Finally, because microcarcinomas are reported to have minimal clinical significance, all the analyses were repeated using size cutoffs of 5 mm and 10 mm, with all smaller tumors coded as benign for the analysis. These alternate results are reported only where they differed significantly from the standard coding. Univariate analysis using 
2 and multivariate analysis using binary logistic regressions were used to determine factors associated with the decision to perform thyroid lobectomy versus total thyroidectomy. A P value of < .05 was accepted as significant.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The procedures were performed by nine surgeons (69% were performed by just three surgeons). Open biopsy alone was performed in 5 patients (1%), isthmusectomy or pyramidal lobe resection in 9 patients (2%), lobectomy with or without isthmusectomy in 285 patients (50%), and subtotal or total thyroidectomy in 265 patients (47%).
Final pathology results revealed carcinoma in 293 patients (52%), with papillary carcinoma the most common diagnosis, seen in 228 patients (78%). Distribution of the histologic types is listed in Tables 1 and 2. The median malignant tumor size was 2 cm (range, 0.04–17.5 cm). For the alternate analysis, in which microcarcinomas were excluded, there were 231 patients with a malignancy over 10 mm in diameter and 262 with a maligancy over 5 mm in diameter on final pathology.
|
|
|
|
|
In patients with suspicious FNA results (n = 144), there were three false-positive (14%) and three false-negative (7%) frozen section reports. A deferred frozen section was reported in 37% of these patients; in this group, the incidence of carcinoma was 51%.
In the 234 patients in whom no acceptable FNA was submitted or performed, there were two false-positive (rate 5%) and eight false-negative (rate 12%) frozen section reports. Of the 82 patients who had neither FNA nor frozen section, 63% had a malignant tumor resected.
Overall, a positive FNA result was reported in 51% of patients biopsied who had a malignancy confirmed on permanent histopathology (Table 2). A negative result was reported in 46% of patients in whom biopsy revealed benign disease on permanent histopathology (Table 1). Tables 3 and 4 display all of the FNA and frozen section results and their association with the final histopathology. These tables are used to calculate sensitivity, specificity, accuracy, and predictive value. In patients who came to operation, the sensitivity/specificity pairs for FNA in the diagnosis of malignancy were 87%/46% for calculation A (suspicious results analyzed with positives) and 51%/99% for calculation B (suspicious results analyzed with negatives), with an accuracy of 68% (calculation A) and 72% (calculation B). The positive predictive value (for malignancy) of a positive FNA was 98% (calculation B). When positive and suspicious results were considered together, the predictive value fell to 66% (calculation A). The negative predictive value of a negative FNA was 74% (calculation A), and when negatives were combined with suspicious results the negative predictive value was 62% (calculation B). The range of sensitivity and specificity for FNA did not change significantly when microcarcinomas were considered benign.
The sensitivity/specificity pairs for frozen section analysis for the diagnosis of carcinoma were 89%/68% for calculation A (deferred results analyzed with positives) and 57%/98% for calculation B (deferred results analyzed with negatives), with an accuracy of 77% and 80%, respectively. The positive predictive value of a positive intraoperative frozen section result was 95% (calculation B), dropping to 69% when positive and deferred results were combined. The negative predictive value of a negative intraoperative frozen section result was 88% (calculation A) and decreased to 74% (calculation B) when deferred results were considered negative. The range of sensitivity and specificity for frozen section did not change significantly when microcarcinomas were considered benign.
The ability of FNA and frozen section to identify the various histologic diagnoses correctly is shown in Tables 1 and 2. Percentages displayed are calculated for number of FNA or FS done. For malignant lesions, the distributions of surgical procedures performed by tumor type are shown in Table 2. The majority of benign thyroid disease was treated by lobectomy with or without isthmusectomy (192 patients [71%]).
Sixty percent of patients aged 20 to 45 and 47% of the remaining patients had a malignant tumor on final pathology (P < .001). There was no difference by age group in percent of patients having total thyroidectomy. Sixty-eight percent of males and 46% of females had a malignant tumor on final pathologic examination (P < .001). Fifty-one percent of males and 46% of females were treated with a total thyroidectomy; however, this difference was not significant (P = NS).
A total thyroidectomy was performed in 72% of patients whose preoperative FNA results were positive for malignancy (P < .001). A lobectomy was performed in 73% of patients with negative FNA results (P < .001). Forty-one percent of all patients with suspicious FNA underwent a total thyroidectomy (P = NS). Forty-three percent of all total thyroidectomies and 40% of all lobectomies were performed without a preoperative FNA recorded (P = NS).
Only 26% of all total thyroidectomies were associated with a positive frozen section result (P < .001) (Table 3). Seventy-six percent of the patients with negative frozen section reports were treated with lobectomy with or without isthmusectomy (P < .001). Eighty-six percent of patients in whom the frozen section was deferred were treated by lobectomy with or without isthmusectomy (P < .001). A total thyroidectomy was performed in 83% of patients in whom no frozen section was obtained (P < .001).
Logistic regression analysis (Table 5) revealed that a positive frozen section result was associated with total thyroidectomy (P < .001, RR 6 [CI 3–10]), whereas a deferred frozen section was associated with lobectomy (P < .05, RR 0.5 [CI 0.3–0.96]). The strongest factor associated with total thyroidectomy was the lack of a frozen section (P < .001, RR 16 [CI 9–27]).
|
Nine patients had procedures terminated at lobectomy or lobectomy with isthmusectomy based on frozen section reports. Of the five patients in this group with a negative frozen section report, four had false-negative reports, and one of these patients went on to completion thyroidectomy. Three patients in this group had deferred frozen section reports that were malignant on final histology, and two of these three went on to completion thyroidectomy. The final patient had a positive frozen section and a small low grade carcinoma, and a completion thyroidectomy was performed at the patient’s request.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Many authors have recommended FNA as an essential step in the evaluation of the solitary thyroid nodule.3–7 Identifying a malignant tumor preoperatively makes operative planning straightforward. Determination of sensitivity and specificity for FNA in this population is limited to patients who came to operation. The prevalence of malignancy in this selected population is higher and will affect the applicability of these results to all patients. In papers that evaluated a similar population for high prevalence of malignancy, the quoted sensitivity ranges from 70% in follicular lesions only6 to 100% in a mixed group,8 and specificity of over 85% is common.7–10 These reported numbers are based on either grouping the suspicious and or follicular results with the positive FNA results11,12 or the negative FNA results,5 or discounting them entirely.9,10,13,28Arranging the suspicious results in this way can yield a seemingly high positive or negative predictive value, accordingly. In the present study, the difficulty of applying these numbers to practice becomes evident. Over 40% of the preoperative FNA resulted in suspicious findings. The predictive value of a positive FNA is 98%; however, when the suspicious results are considered positive as well, the predictive value drops to 66%. Twenty-six percent of patients with negative FNA results in this cohort had malignancy on final pathology. Of greatest concern was that in patients with a proven malignancy, only 51% of those tested had a positive FNA. Although 87% of these patients had positive or suspicious FNA results, 13% were reported as benign preoperatively. The false-negative rate remained at 9% to 10% when the analysis was repeated using a 5- or 10-mm cutoff for microcarcinomas. A negative FNA should not be used as reassurance if other clinical factors are suggestive of malignancy. From this analysis it is evident that patients with negative or suspicious FNA results were brought to the operating room based on factors other than these results.
The technique of FNA and the intent of the procedure were not specifically evaluated in this study. Missing the diagnosis in a solitary lesion clearly is not the same as missing a small cancer in a multinodular gland. Conversely, identifying an incidental microcarcinoma during FNA may lead to a resection that has no effect on clinical course. The alternate analysis of the data using size cutoffs for microcarcinomas illustrates this difficulty well: whereas a larger cutoff size yields a lower false-negative rate, the false-positive rate increases. Overall, the measures of accuracy do not change, so the clinician must make a decision based on the available information.
There are several excellent large-scale studies delineating the risk factors for malignancy in a thyroid tumor.21–23 History of radiation exposure, lack of response to thyroid suppression, local symptoms including recurrent nerve compression or dysphagia, evidence of metastatic disease, and tumor size have been implicated as well. These factors may be identified preoperatively, providing an indication for operation and an operative plan. Although the present study did not analyze these factors specifically, we have reported previously on the importance of risk stratification in the management of thyroid tumors at our institution.24–27
Once the operation has begun there are some further clues that can be obtained to identify a thyroid malignancy and determine the patient’s subsequent risk. On exploration, the presence of lymph nodes or soft tissue masses in the operative field may indicate metastatic disease or local invasion and a frozen section of this tissue would be relatively straightforward for the diagnosis of malignancy. Palpation of the entire thyroid gland at exploration may discover other lesions, which may indicate bilateral disease. Finally, a frozen section of the thyroid gland itself could document the existence of a malignancy.
Many studies of the utility of frozen section in thyroid surgery have been performed.5,11,13,29–31 Again, the sensitivities and specificities reported often are over 80%.8–10,19 Many of these calculations are performed without including the deferred or follicular groups. The true sensitivity and specificity in the present study lie within the range reported. When one compares the 46% of deferred patients who had a malignant tumor on final histology to the 70% malignancy rate in patients who had no frozen section, there is a tendency to question the utility of frozen section at all. Some authors report improved sensitivity and specificity for the detection of malignancy when FNA and FS are combined, and they recommend FS be used in every case,5,11,13 whereas others have outlined the indications for a frozen section in an effort to improve its utility in thyroid surgery.7,9,12,14–17 They recommend frozen section to clarify a suspicious, unsatisfactory, or absent FNA result. When applied to these groups in the present study, (Fig. 1), 36% are deferred, and there is a 10% false-negative rate and a 6.5% false-positive rate. These numbers do not significantly improve the rate of detection of a malignancy over FNA and clinical factors alone. Our results are not unique. Several groups have reported similar numbers. Some have concluded that frozen section is helpful in these patients,7,9,12,14–17 whereas others report that frozen section adds little to the management of these lesions.3,8,10,18–20 We did not consider the use of frozen section for the evaluation of locally advanced or metastatic disease in the present study, but it seems clear that in this selected population frozen section has a significant effect on operative management.
This study differs from the majority of the studies listed above in that we have included patients with both benign and malignant diagnoses. Conclusions about predictive factors that are based only on cohorts of patients with malignant disease may not apply to patients with benign disease. In the present study the prevalence of malignancy was 52%, allowing us to base our conclusions on a more realistic situation. As we have just demonstrated, it is not possible to know 100% of the time that a thyroid tumor is malignant until the final pathology report is received.
The results of the multivariate logistic regression confirm that at our institution the majority of total thyroidectomies are performed for reasons other than a positive FNA. A positive frozen section was associated with a total thyroidectomy; however, an even stronger predictor was the lack of a frozen section. These were patients in whom the indications for total thyroidectomy by history, physical examination, risk factors, and or studies were strong enough that an attempt at intraoperative confirmation of the diagnosis was not warranted. This includes the 29% of patients with benign thyroid disease who underwent total thyroidectomy for bilateral lesions.
Although 18% of patients had a positive frozen section result, chart review revealed that in only 5% of cases was the extent of resection actually altered by the frozen section results. Even with a 12% false-negative rate for frozen sections, only three patients were returned to the operating room for completion thyroidectomy. Our view differs from those published by other high-volume centers.5,7,11,13,32 We do not advocate total thyroidectomy for every thyroid cancer. Each case is decided individually based on risk factors and intraoperative findings. For this reason, the majority of decisions about the extent of resection in thyroid cancer can be made without relying on an FNA or frozen section result.
Most of the errors in FNA are related to sampling error, so the percentage of suspicious and false-negative FNA results could be reduced in this cohort by improved technique. In addition, the false-negative and deferred diagnosis rates for frozen section could be improved by increasing the number of sections examined intraoperatively.5 The benefit of improving the performance of these tests is questionable, however, because they are outweighed by other prognostic factors in our management algorithm.
The data presented in this study support the use of FNA in the confirmation of malignancy and the need for operation. However, FNA is not very sensitive in practice, and a negative FNA should not prevent an operation that is indicated by other clinical factors. The extent of resection should be individualized based on the established risk factors in combination with findings at neck exploration. The routine use of frozen section analysis is not warranted, because it adds little to this decision-making process.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received for publication March 16, 2000. Accepted for publication October 2, 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  G. F. Huber, P. Dziegielewski, T. W. Matthews, S. J. Warshawski, L. M. Kmet, P. Faris, M. Khalil, and J. C. Dort Intraoperative Frozen-Section Analysis for Thyroid Nodules: A Step Toward Clarity or Confusion? Arch Otolaryngol Head Neck Surg, September 1, 2007; 133(9): 874 - 881. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T.-C. Chao, J.-D. Lin, H.-H. Chao, C. Hsueh, and M.-F. Chen Surgical Treatment of Solitary Thyroid Nodules Via Fine-Needle Aspiration Biopsy and Frozen-Section Analysis Ann. Surg. Oncol., February 1, 2007; 14(2): 712 - 718. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Y. Greenblatt, T. Woltman, J. Harter, J. Starling, E. Mack, and H. Chen Fine-Needle Aspiration Optimizes Surgical Management in Patients With Thyroid Cancer Ann. Surg. Oncol., June 1, 2006; 13(6): 859 - 863. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Stojadinovic, S. I. Fields, C. D. Shriver, S. Lenington, R. Ginor, G. E. Peoples, H. B. Burch, T. Peretz, H. R. Freund, and A. Nissan Electrical Impedance Scanning of Thyroid Nodules Before Thyroid Surgery: A Prospective Study Ann. Surg. Oncol., February 1, 2005; 12(2): 152 - 160. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. B. Umbricht, G. T. Conrad, D. P. Clark, W. H. Westra, D. C. Smith, M. Zahurak, M. Saji, R. C. Smallridge, S. Goodman, and M. A. Zeiger Human Telomerase Reverse Transcriptase Gene Expression and the Surgical Management of Suspicious Thyroid Tumors Clin. Cancer Res., September 1, 2004; 10(17): 5762 - 5768. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. A. Callcut, S. M. Selvaggi, E. Mack, O. Ozgul, T. Warner, and H. Chen The Utility of Frozen Section Evaluation for Follicular Thyroid Lesions Ann. Surg. Oncol., January 1, 2004; 11(1): 94 - 98. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
