Clinical Endocrinology (1999) 51, 509–515

509q 1999 Blackwell Science Ltd

Sensitivity and specificity of the fine needle aspirationbiopsy of the thyroid: clinical point of view

J. Cap*,† A. Rys ka‡, P. Rehor kova ,† E. Hovorkova´‡,Z. Kerekes‡ and D. Pohneˇtalova ‡*First Medical Clinic, †Second Medical Clinic and‡Department of Pathology, University Hospital, CharlesUniversity, Hradec Kralove, Czech Republic

(Received 18 January 1999; returned for revision 12 May 1999;finally revised 2 June 1999; accepted 27 June 1999)

Summary

INTRODUCTION The rates of sensitivity and specifi-city of fine needle aspiration biopsy (FNAB) for thediagnosis of thyroid malignancy differ considerablyamong various reported series. These values areinfluenced by three factors: (a) whether only clearlypositive and negative results are considered, orwhether the commonly encountered 10–20% of inde-terminate/suspicious ones are included; (b) whetheradenomas are considered as neoplasms in one groupwith carcinomas; and (c) whether only histologicallyproven cases are used in calculations or whetherpatients with benign clinical follow-up are included.AIM The aim of the study was to evaluate the sensi-tivity and specificity of FNABs performed at thisinstitution in the last 7 years from the clinical pointof view, considering only benign vs. suspicious/malignant FNAB results (indicating surgery), andbenign (including adenomas) vs. malignant definitivehistology.STUDY DESIGN Retrospective study comparing pre-operative FNAB results with definitive histologicalexamination after operation.PATIENTS A total of 2492 FNABs were performed in2100 patients (1875 women and 225 men); their agesranged from 9 to 85 years, with a median of 46 years.Clinical diagnosis was multinodular goitre in 1330,single nodule in 591, Hashimoto’s thyroiditis in 147and subacute thyroiditis in 32 cases. In 148 instances,the nodule was cystic. A history of previous treatmentfor carcinoma of the thyroid was present in 12patients. Five hundred and thirty-six patients subse-quently underwent thyroid surgery.

STATISTICS The values of sensitivity, specificity,positive predictive value (PPV), negative predictivevalue (NPV) and diagnostic accuracy were calculated.RESULTS The sensitivity was 86%, specificity 74%,PPV 34%, NPV 97% and diagnostic accuracy 75%.CONCLUSIONS The specificity and positive predic-tive value are low when fine needle aspirationbiopsy results are divided into two categories only(these being indication for surgery or not), and whenonly suspicious/malignant fine needle aspirationbiopsies with subsequent malignant histology areconsidered to be true positive. Nevertheless, the abil-ity to discriminate 11·7% of patients with a 34% prob-ability of malignancy (suspicious/malignant cytology)from 81·2% of patients (benign cytology) with a prob-ability of only 3% is very helpful.

Fine needle aspiration biopsy (FNAB) is a well establishedprocedure in the primary diagnosis of thyroid disorders(Gutman & Henry, 1998). This technique has resulted in anoverall decline in the number of thyroidectomies performed forthyroid nodules (Julianet al., 1989; Carpiet al., 1996) and aconcurrent increase in the diagnosis of thyroid neoplasms(Gallowayet al., 1991; Gharibet al., 1993). Specific limitationsof this technique that result in false-negative and false-positiveresults are well known (Galera-Davidson, 1997); a substantialproportion of FNAB results are neither clearly benign norclearly malignant and fall into the category of indeterminate/suspicious results. The calculations of sensitivity and specificityof FNAB for the diagnosis of thyroid malignancy differconsiderably among individual series in the range of 65–98%and 72–100%, respectively (Gharib & Groellner, 1993). Theaim of this retrospective study was to correlate FNAB withhistological results to evaluate the sensitivity and specificity ofthis diagnostic procedure for pre-operative detection ofmalignancy. As thyroid surgery is indicated in patients withboth malignant and indeterminate/suspicious FNAB, and theindication for surgery represents the main clinical impact of theFNAB result, only two categories of diagnostic FNABs wereconsidered: suspicious and non-suspicious. Suspicious FNABresults were considered as true positive only in cases withhistologically proven malignancy and not in cases with a finaldiagnosis of follicular adenoma as in the latter, neither theextent of thyroid surgery, nor further treatment, differed fromthat of a hyperplastic nodule of colloid goitre.

Correspondence: Dr Jan Cˇ ap, Charles University, University Hospital,Second Medical Clinic, CZ-500 05 Hradec Kra´love, Czech Republic.Fax:þ 420 49 5832003; E-mail: jan.cap@telecom.cz

Subjects, materials and methods

Patients

During a 7-year period (1991–98), a total of 2492 fine needleaspiration cytologies were performed in 2100 patients. Theaspiration was repeated in more than 200 patients (up to fourtimes in three cases). Women predominated (88%) and agesranged from nine to 85 years, with the median being 46 years.The clinical indication for FNAB was multinodular goitre inmost cases (1330); a single nodule was present in 591 cases.Hashimoto’s thyroiditis was suspected in 147 patients andsubacute thyroiditis in 32. In 148 instances, the nodule wascystic. A history of previous treatment for carcinoma of thethyroid was present in 12 patients.

Some 536 patients had subsequent thyroid surgery and thedefinitive histology was available. The indication for surgerywas either compressive symptoms or a suspicious FNABcytological finding.

Follow-up of at least 2 years (2–7 years with the median of4·2 years) after the FNAB was available in 773 patients who didnot have an operation (including 17 with suspicious FNAB).Most of these patients were treated with suppressive doses ofthyroxine and in all of them, there was a decrease in size, or atleast no progression of goitre volume, and no signs ofmalignancy.

Cytology

The aspirations were performed by an endocrinologist using a25-gauge needle attached to a 20-ml syringe. The number ofpasses was dependent on the size of the nodule and the amountof material obtained with each pass. Four passes were done inthe majority of cases. The smears were air-dried and stainedwithout fixation using May-Gru¨nwald-Giemsa staining; thenumber of smears ranged from one to 10 with a median of fourslides per case.

For the purpose of this study, the results of the cytologicalevaluation were classified as:X inadequate (less than 10 groups of cells, each containing at

least 10 elements)X benign (including nodular goitre with or without regressive

changes and/or focal lymphocytic thyroiditis, Hashimoto’sand deQuervain’s subacute thyroiditis)

X suspicious of malignancy/malignant (including follicularneoplasia, papillary carcinoma, anaplastic carcinoma,medullary carcinoma, lymphoma and metastatic carcinoma;these were diagnosed either unequivocally or with varyingdegrees of probability, and include uncertain findings thatcould not rule out malignancy, classified as ’others’).The results were categorized on the basis of the findings of

the cytopathologist before the operation.

Most FNABs were performed in a clinic after palpation of thedominant nodule(s). Only 25% were done under ultrasoundguidance. This was used in patients with small nodules or incases where the first cytology had been inadequate.

Histology

Histology included surgical excisions of lesions that were thetarget of cytological evaluation. These were classified as:X benign (colloid nodular goitre, follicular adenoma, Hashi-

moto’s thyroiditis)X malignant (follicular carcinoma, papillary carcinoma, ana-

plastic carcinoma, medullary carcinoma, lymphoma ormetastatic carcinoma).

Statistical analysis

The numbers of true-positive (TP), true-negative (TN), false-positive (FP) and false-negative (FN) results were calculated.The suspicious/malignant FNABs were considered as truepositives (TP) in cases where operation revealed a malignancyon histological examination, and they were considered falsepositives (FP) when no malignancy was found. The benignFNAB was considered as a true negative (TN) if the histologicalfinding was benign and false negative (FN) in cases ofhistologically proven malignancy.

From these numbers, the following statistical values havebeen calculated:X sensitivity in percentage: (TP/TPþ FN) * 100;X specificity in percentage: (1-(FP/FPþ TN))(TN/TN þ FP) *

100;X positive predictive value (PPV) in percentage: (TP/

TPþ FP)* 100;X negative predictive value (NPV) in percentage: (TN/

TN þ FN) * 100;X diagnostic accuracy in percentage: (TPþ TN/

TPþ TN þ FPþ FN) * 100.

Ethical notes

The decision to perform both the FNAB and thyroid surgerywas based on clinical indications only and could not have beeninfluenced by this retrospective evaluation. Only appropriatephysicians reviewed the records, and the individual patient dataremained anonymous.

Results

A total of 187 examinations in 150 patients was inadequate fordiagnosis (7·5% of the total number of FNABs). Twenty ofthese patients had subsequent surgery and one had medullary

510 J. Cap et al.

q 1999 Blackwell Science Ltd,Clinical Endocrinology, 51, 509–515

carcinoma. The histological diagnoses of the others werebenign goitre. They are not included in any further evaluations.

The diagnostic results of FNABs in 1950 patients aresummarized in Table 1. It was benign in 1705 (81·2%) casesand suspicious/malignant in 245 (11·7%). Thyroid surgery wasperformed in 339 of 1750 (19·5%) patients with benign FNABresults and in 177 of 245 (72·2%) cases with suspiciouscytological classifications. It is evident that the suspiciousFNAB result and the need for histological evaluation of thenodule were the main reasons for operation in many patients.

The final histological diagnosis was malignant in 10 lesionsclassified cytologically as non-suspicious (false-negative

results). The two lymphomas were misinterpreted as Hashimo-to’s thyroiditis. Of the eight carcinomas, five were papillarymicrocarcinomas of 10 mm and less, found in large multi-nodular goitres. The evaluated cytological specimen was notobtained from these small lesions (sampling error). In one case,the papillary carcinoma showed cystic degeneration, withresidual tumour nests in the wall of the cyst resulting in afalse negative interpretation as benign regressive changes. Inthe two remaining cases, the papillary carcinoma wasmisinterpreted by the cytopathologist as a benign nodule, andthe smears were classified as suspicious on retrospective re-evaluation.

Thyroidal fine needle aspirations 511

q 1999 Blackwell Science Ltd,Clinical Endocrinology, 51, 509–515

Table 1 Results of FNAB and histological diagnosis in 1950 patients

Histological classificationNumber

Cytological Number of of patients Nodular Hashimoto’sclassification patients operated goitre/adenoma thyroiditis Carcinoma Lymphoma

Benign 1705 339 322 7 8 2Colloid goitre 1153 266 256 3 7Regressive 302 66 64 1 1changes/cystHashimoto’s 230 7 2 3 2thyroiditisDeQuervain’s 20 0thyroiditis

Suspicious 245 177 105 12 56 4Anaplastic 6 6 6Follicular 107 70 55 1 14Papillary 47 41 16 1 24Medullary 5 5 1 4Lymphoma 7 6 2 4Hurthle cell 61 45 30 8 7Others 12 4 3 1

Table 2 Numbers of true-positive (TP), false-positive (FP), true-negative (TN) and false-negative (FN) results of FNA cytology and correspondingvalues of sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and diagnostic accuracy (DA) using four differentapproaches

No. of Sensitivity Specificity PPV NPV DApatients TP FP TN FN (%) (%) (%) (%) (%)

Only histologically verified cases, follicular adenoma is false-positive516 60 117 329 10 86 74 34 97 75

Only histologically verified cases, follicular adenoma is true-positive516 91 86 329 10 90 79 51 97 81

Follow-up cases included, follicular adenoma is false-positive1289 60 135 1084 10 86 89 31 99 89Follow-up cases included, follicular adenoma is true-positive1289 91 104 1084 10 90 91 47 99 91

Among 177 operated patients with a suspicious finding onaspiration cytology, only 60 malignancies (56 carcinomas andfour lymphomas) were found. There were 117 false-positiveresults. The positive predictive value of suspicious FNABdepended on tumour type. It was 6/6 (100%) for anaplastic, 4/5(80%) for medullary, 24/41 (58·5%) for papillary carcinomaand 4/6 (66·7%) for lymphoma, but only 14/70 (20%) forfollicular neoplasia and 7/45 (15·6%) for Hu¨rthle cell tumour.In 31 cases classified cytologically as follicular neoplasia, thefinal histological diagnosis was follicular adenoma.

In histologically verified cases, the results of FNA cytologywere false negative (FN) in 10 cases, false positive (FP) in 117cases, true negative (TN) in 329 and true positive (TP) in 60cases. Using a different approach, and considering the result ofFNA cytology in the 31 cases of follicular neoplasia with thehistological diagnosis of follicular adenoma as true positive, thenumbers are TP¼ 91 and FP¼ 86.

In all, 773 patients were followed up for more than 2 yearsafter the FNA cytology was performed without any clinical signof thyroid tumour. If these patients are considered to be withoutmalignancy, the number of true-negative cases increasesdramatically, with a small increase in the number of false-positive cases as well. The calculations of sensitivity,specificity, positive and negative predictive values anddiagnostic accuracy for these various approaches are summar-ized in Table 2.

Discussion

Fine needle aspiration biopsy has been generally accepted as arapid, accurate and inexpensive means of evaluating thyroidswellings and is currently in widespread use, especially fordiagnosing malignancy among thyroid nodules (Giuffrida &Gharib, 1995; Gutman & Henry, 1998). FNAB has two majorlimitations: inadequate results and suspicious or indeterminateresults. Only smears containing five to six (at some institutionseight or 10) groups of well preserved cells, with each groupcontaining at least 10–15 cells, are considered satisfactory(Gutman & Henry, 1998). On the other hand, Hamburger(1994) suggested that limited numbers of benign cells andabundant colloid may be sufficient for observation. Inadequateor unsatisfactory results are obtained when lesions are cystic orvascular, yielding a dilute specimen and few follicular cells.Their frequency depends on both the skill of the personperforming aspiration and the stringency of the cytopatholo-gist’s criteria, varying between 2 and 21% in different series(Gharib & Goellner, 1993). In the material studied here, 187smears (7·5%) obtained from 150 patients were inadequate fordiagnosis. Twenty of these patients had subsequent surgery andthe final histological diagnosis was benign goitre in all but oneman who had a history of thyroidectomy and radiation therapy

for medullary carcinoma. He had a clinically evident localrelapse which was subsequently histologically proven to bemedullary cancer, but the FNAB specimen was non-diagnostic.However, no carcinoma was missed clinically, and one of 20operated cases (5%) was malignant following a non-diagnosticsmear. This is in accordance with previous reports whichdescribe the frequency of carcinomas to be 2% (MacDonald &Yazdi, 1996), 5% (Schmidtet al., 1997) and even 9%(McHenry et al., 1993) of operated patients with inadequateFNABs.

Diagnostic FNAB smears are traditionally categorized asbenign, malignant (unequivocal malignancy in the case ofanaplastic, medullary and papillary carcinoma of the thyroid,metastatic carcinoma and lymphoma), and suspicious/indeter-minate (follicular neoplasia, Hu¨rthle cell tumour, suspicious butnot diagnostic for papillary carcinoma, or other suspiciousfindings) (Gutman & Henry, 1998). There is certainly no cleardistinction between unequivocally malignant cytology and amore or less suspicious one. Hamburger (1994) categorizedFNAB diagnoses as probably malignant, possibly malignant,and benign. Others differentiate between benign, equivocal,suspicious and malignant FNABs (Klemiet al., 1991).Follicular and Hu¨rthle cell lesions can be subdivided intotype I (probable carcinoma), type II (highly suspicious lesion)and type III (probable adenoma) (Willems & Lo¨whagen, 1981).In the largest series, suspicious findings account for between11% (Cersosimoet al., 1993; Gharib & Goellner, 1993) and17% (LaRosaet al., 1991), with only 3–4% malignant results.In some series, however, the proportion of malignant FNABs ishigher than that of suspicious ones (Balochet al., 1998). Thismay reflect not only differences in cytopathological classifica-tion, but also populations studied with higher rates of malignantresults in series where only clinically suspicious nodules areaspirated. Most of the suspicious findings represent follicularand Hurthle cell tumours, where the distinction betweenadenoma and carcinoma depends on the demonstration ofinvasion of the capsule and/or angioinvasion, features notrecognizable at the cytological level. From the clinical point ofview, therefore, both malignant and suspicious FNABsrepresent an indication for thyroid surgery, and about 20% ofnodules with suspicious FNAB are malignant (La Rosaet al.,1991; Gharib & Goellner, 1993; Schlinkertet al., 1997). At thisinstitution, intra-operative frozen sections are used before totalthyroidectomy in patients with malignant and suspiciouspapillary FNAB, where the probability of a confident intra-operative diagnosis is high (Gibb & Pasieka, 1995). In noduleswith suspicious follicular or Hu¨rthle cell FNAB, lobectomy isperformed; further treatment depends on the result of finalparaffin section histology as we, as well as others (Bronneret al., 1994), do not consider frozen sections useful in this typeof lesion.

512 J. Cap et al.

q 1999 Blackwell Science Ltd,Clinical Endocrinology, 51, 509–515

The major clinical impact of FNAB is the indication forsurgery. In this study, malignant and suspicious results weretherefore evaluated as a single group, leaving only twodiagnostic categories of FNABs: benign and suspicious/malignant. Of a total 2492 results, 257 (10·3%) in 245 patientswere malignant/suspicious and 2048 (82·2%) performed in1705 patients were benign.

The rates of sensitivity, specificity, positive predictive value(PPV) and negative predictive value (NPV) differ considerablyamong various series (see Table 3).

These differences have the following three main sources.(1) Only benign and clearly malignant results are evaluated

in series with high sensitivity and PPV, omitting a considerableproportion of suspicious/indeterminate results from the calcu-lations. If the statistics of one of the largest series (La Rosaet al., 1991) are recalculated considering suspicious follicularFNABs as positive, the specificity drops from 98 to 46% andPPV, from 98·3 to 44%.

(2) While, in the majority of series, only cases with definitivehistology are calculated, some consider patients with negativefollow-up as negative for malignancy (Leonard & Melcher,1997; Balochet al., 1998). This approach increases the numberof truly negative cases, with an increase in specificity and NPV.We feel that malignancy cannot be ruled out with certaintywithout histological evaluation, as differentiated carcinomaneed not progress for years (Ain, 1995).

(3) The statistical values depend on whether the adenomasare categorized as neoplasms in one group with cancers (andsuspicious FNABs are then considered as truly positive) or asbenign lesions. While the first approach reflects the impossi-bility of distinguishing follicular and Hu¨rthle cell adenomasfrom their differentiated carcinoma counterparts cytologically,from the clinical point of view, the adenoma is a benign lesionand neither radical thyroidectomy nor radioactive iodinetherapy are indicated.

When evaluating histologically proven cases only andconsidering adenomas as benign, the sensitivity of FNAB forthe diagnosis of malignancy in the present group of patients hasbeen 86%, specificity 74%, diagnostic confidence 75%, positivepredictive value 34% and negative predictive value 97%. Whilethe NPV is comparable with the results of previous series, thePPV using this approach is rather low. It differs considerablydepending on the type of suspicious/malignant result. Thepositive predictive value was 100% for anaplastic carcinomaand high for medullary (80%) and papillary carcinoma (58·5%)as well as for lymphoma (66·8%), but it was only 20% forfollicular neoplasia and 15·6% for Hu¨rthle cell tumour. Theserates are comparable with those reported in previous series,where the PPV for follicular neoplasia and Hu¨rthle cell tumourvaried from 12% (Schlinkertet al., 1997) to 50% (Gibb &Pasieka, 1995). In our series, the somewhat higher PPV forfollicular cell lesions than for Hu¨rthle cell lesions is unusual, asmost of the previous series were the opposite (Cersosimoet al.,1993; Tyleret al., 1994; Gibb & Pasieka, 1995).

From the clinical point of view, the most troublesome are thefalsely-negative results, as the appropriate treatment might bedelayed. In our series, eight carcinomas were missed by FNAB;five of them were small malignancies (four papillary micro-carcinomas of 5, 8, 10 and 10 mm in diameter, and a 12-mmmetastasis of renal cell carcinoma to the goitre). It is thoughtthat these lesions were not sampled during aspiration. Onlyabout 25% of FNABs in our patients were performed under thecontrol of ultrasound, especially in cases of small nodules or aninadequate sample from previous routine (clinic) FNAB (Rosenet al., 1993). However, it is thought that in larger multinodulargoitres, small lesions can be missed even with the use ofultrasound guidance. None of the four missed microcarcinomashas been re-operated or treated with radioiodine. One papillarycarcinoma was not recognized because of its cystic degenera-tion, which is a situation described in the literature (Meko &

Thyroidal fine needle aspirations 513

q 1999 Blackwell Science Ltd,Clinical Endocrinology, 51, 509–515

Table 3 Reported values of sensitivity,specificity, positive predictive value (PPV) andnegative predictive value (NPV) in previousseries

Series Sensitivity (%) Specificity (%) PPV (%) NPV (%)

Schnurer & Widstrom, 1978 96 76 100 99·5Lowhagenet al., 1979 91 69 100 96Gardineret al., 1986 65 91 100 88·5Abu-Nemaet al., 1987 88 100 100 98·9Hawkinset al., 1987 86 95 95·4 97·6Hall et al., 1989 84 90 98·7 97La Rosaet al., 1991 98 98 98·3 97·3Gharib & Goellner, 1993 98 99 98 99·3Hollemanet al., 1995 84 52 53 83Leonard & Melcher, 1997 88 78 46 97Hamminget al., 1998 67 99 96 88Balochet al., 1998 92 84 73·3 98·7

Norton, 1995). The remaining two papillary carcinomas(follicular variant of papillary carcinoma and papillarycarcinoma with regressive changes) were misinterpreted bythe cytopathologist as benign hyperplastic nodules.

In seven cases of papillary carcinomas, the FNA results wereclassified as follicular neoplasia. Among these, only one tumourfulfilled the established criteria of follicular variant of papillarycarcinoma on histopathological examination. In two cases theproportion of the follicular histological pattern exceeded 66%,was 50% in two others and less than 30% in the remaining two.Papillary carcinomas are composed of a mixture of neoplasticfollicles and papillae, and the cytological result depends on theproportion of follicular component not only in the tumour as awhole but especially in the site from which the cytologicalspecimens have been obtained (Gallagheret al., 1997). Fromthe clinical point of view, both types of malignant/suspiciousFNA results indicate surgery.

Two lymphomas were incorrectly interpreted as Hashimoto’sthyroiditis. The distinction of these two conditions is proble-matic (Goellner, 1997), and the detection of B-cell clonalitymay be helpful (Lovchiket al., 1997). In both cases, surgerywas indicated because of a lack of response to appropriateconservative treatment. Repeated FNAB revealed lymphoma inone of these female patients.

The proportion of false positive results in patients withFNAB findings that indicated surgery was rather high. Never-theless, this diagnostic procedure was able to distinguishpatients with a 34% probability of malignancy (suspicious/malignant cytology, 11·7% of patients) from those (benigncytology, 81·2% of patients) with a probability of only 3%.

Acknowledgements

The authors thank Dr Preston Simpson (Department ofPathology, St Agnes Medical Center, 1303 East HerndonAve., Fresno, CA 93720), who kindly read the manuscript andoffered several helpful suggestions.

References

Abu-Nema, T., Ayyash, K. & Tibblin, S. (1987) Role of aspirationbiopsy cytology in the diagnosis of cold solitary thyroid nodules.British Journal of Surgery, 74, 203.

Ain, K.B. (1995) Papillary thyroid carcinoma. Etiology, assessment,and therapy.Endocrinology and Metabolism Clinics of NorthAmerica, 24, 711–760.

Baloch, Z.W., Sack, M.J., Yu, G.H., LiVolsi, V.A. & Gupta, P.K.(1998) Fine-needle aspiration of thyroid: an institutional experience.Thyroid, 8, 565–569.

Bronner, M.P., Hamilton, R. & LiVolsi, V.A. (1994) Utility of frozensection analysis on follicular lesions of the thyroid.EndocrinePathology, 5, 154–161.

Carpi, A., Ferri, E., Toni, M.G., Sagripanti, A., Nicolini, A. & DiCoscio, G. (1996) Needle aspiration techniques in preoperativeselection of patients with thyroid nodules: a long-term study.Journalof Clinical Oncology, 14, 1704–1712.

Cersosimo, E., Gharib, H., Suman, V.J. & Goellner, J.R. (1993)‘Suspicious’ thyroid cytological findings: outcome in patientswithout immediate surgical treatment.Mayo Clinic Proceedings,68, 343–348.

Galera-Davidson, H. (1997) Diagnostic problems in thyroid FNAs.Diagnostic Cytopathology, 17, 422–428.

Gallagher, J., Oertel, Y.C. & Oertel, J.E. (1997) Follicular variant ofpapillary carcinoma of the thyroid: fine-needle aspirates withhistologic correlation.Diagnostic Cytopathology, 16, 207–213.

Galloway, J.W., Sardi, A., DeConti, R.W., Mitchell, W.T. Jr & Bolton,J.S. (1991) Changing trends in thyroid surgery. 38 years’ experience.American Surgery, 57, 18–20.

Gardiner, G.W., de Souza, F.M., Carydis, B. & Seemann, C. (1986)Fine-needle aspiration biopsy of the thyroid gland: results of a five-year experience and discussion of its clinical limitations.Journal ofOtolaryngology, 15, 161–165.

Gharib, H. & Goellner, J.R. (1993) Fine-needle aspiration biopsy of thethyroid: an appraisal.Annals of Internal Medicine, 118,282–289.

Gharib, H., Goellner, J.R. & Johnson, D.A. (1993) Fine-needleaspiration cytology of the thyroid. A 12-year experience with11,000 biopsies.Clinics in Laboratory Medicine, 13, 699–709.

Gibb, G.K. & Pasieka, J.L. (1995) Assessing the need for frozen sections:still a valuable tool in thyroid surgery.Surgery, 118,1005–1009.

Giuffrida, D. & Gharib, H. (1995) Controversies in the management ofcold, hot and occult thyroid nodules.American Journal of Medicine,99, 642–650.

Goellner, J.R. (1997) Problems and pitfalls in thyroid cytology.Monographs in Pathology, 39, 75–93.

Gutman, P.D. & Henry, M. (1998) Fine needle aspiration cytology ofthe thyroid.Clinics in Laboratory Medicine, 18, 461–482.

Hall, T.L., Layfield, L.J., Philippe, A. & Rosenthal, D.L. (1989) Sourcesof diagnostic error in fine needle aspiration of the thyroid.Cancer,63, 718–725.

Hamburger, J.I. (1994) Diagnosis of thyroid nodules by fine needlebiopsy: use and abuse.Journal of Clinical Endocrinology andMetabolism, 79, 335–339.

Hamming, J.F., Vriens, M.R., Goslings, B.M., Songun, I., Fleuren, G.J.& van de Velde, C.J. (1998) Role of fine-needle aspiration biopsy andfrozen section examination in determining the extent of thyroidect-omy. World Journal of Surgery, 22, 575–579.

Hawkins, F., Bellido, D., Bernal, C., Rogopoulou, D., Ruiz Valdepenas,M.P., Laronzo, E., Perez-Barrios, A. & De Agustin, P. (1987) Fineneedle aspiration biopsy in the diagnosis of thyroid cancer andthyroid disease.Cancer, 59, 1206–1209.

Holleman, F., Hoekstra, J.B. & Ruitenberg, H.M. (1995) Evaluation offine needle aspiration (FNA) cytology in the diagnosis of thyroidnodules.Cytopathology, 6, 186–175.

Julian, J.S., Pittman, C.E., Accettullo, L., Berg, T.A. & Albertson, D.A.(1989) Does fine-needle aspiration biopsy really spare patientsthyroidectomy?American Surgery, 55, 238–242.

Klemi, P.J., Joensuu, H. & Nylamo, E. (1991) Fine needle aspirationbiopsy in the diagnosis of thyroid nodules.Acta Cytologica, 35,343–438.

La Rosa, G.L., Belfiore, A., Giuffrida, D., Sicurella, C., Ippolito, O.,Russo, G. & Verneri, R. (1991) Evaluation of the fine needleaspiration biopsy in the preoperative selection of cold thyroidnodules.Cancer, 67, 2137–2141.

514 J. Cap et al.

q 1999 Blackwell Science Ltd,Clinical Endocrinology, 51, 509–515

Leonard, N. & Melcher, D.H. (1997) To operate or not to operate? Thevalue of fine needle aspiration cytology in the assessment of thyroidswellings.Journal of Clinical Pathology, 50, 941–943.

Lovchik, J., Lane, M.A. & Clark, D.P. (1997) Polymerase chainreaction-based detection of B-cell clonality in the fine needleaspiration biopsy of a thyroid mucosa-associated lymphoid tissue(MALT) lymphoma.Human Pathology, 28, 989–992.

Lowhagen, T., Granberg, P.O., Lundell, G., Skinnari, P., Sundholm, R.& Williams, J.S. (1979) Aspiration biopsy cytology (ABC) innodules of the thyroid suspected to be malignant.Surgical Clinics ofNorth America, 59, 3–18.

MacDonald, L. & Yazdi, H.M. (1996) Nondiagnostic fine needleaspiration biopsy of the thyroid gland: a diagnostic dilemma.ActaCytologica, 40, 423–428.

McHenry, C.R., Walfish, P.G. & Rosen, I.B. (1993) Non-diagnostic fineneedle aspiration biopsy: a dilemma in management of nodularthyroid disease.American Surgery, 59, 415–419.

Meko, J.B. & Norton, J.A. (1995) Large cystic/solid thyroid nodules:a potential false-negative fine-needle aspiration.Surgery, 118,996–1003.

Rosen, I.B., Azadian, A., Walfish, P.G., Salem, S., Lansdown, E. &

Bedard, Y.C. (1993) Ultrasound-guided fine-needle aspiration biopsyin the management of thyroid disease.American Journal of Surgery,166,346–349.

Schlinkert, R.T., van Heerden, J.A., Goellner, J.R., Gharib, H., Smith,S.L., Rosales, R.F. & Weaver, A.L. (1997) Factors that predictmalignant thyroid lesions when fine-needle aspiration is‘suspicious for follicular neoplasm’.Mayo Clinic Proceedings, 72,913–916.

Schmidt, T., Riggs, M.W. & Speights, V.O. Jr. (1997) Significance ofnondiagnostic fine-needle aspiration of the thyroid.Southern MedicalJournal, 90, 1183–1186.

Schnurer, L.B. & Widstrom, A. (1978) Fine-needle biopsy of thethyroid gland: a cytohistological comparison in cases of goiter.Annals of Otology Rhinology and Laryngology, 87, 224–227.

Tyler, D.S., Winchester, D.J., Caraway, N.P., Hickey, R.C. & Evans,D.B. (1994) Indeterminate fine-needle aspiration biopsy of thethyroid: identification of subgroups at high risk for invasivecarcinoma.Surgery, 116,1054–1060.

Willems, J.S. & Lowhagen, T. (1981) Fine needle aspiration cytologyin thyroid disease.Clinical Endocrinology and Metabolism, 10,247–266.

Thyroidal fine needle aspirations 515

q 1999 Blackwell Science Ltd,Clinical Endocrinology, 51, 509–515