www.elsevier.com/locate/jpedsurg

Journal of Pediatric Surgery (2009) 44, 2184–2191

Reviews

Pediatric thyroid fine-needle aspiration cytology:a meta-analysisChristopher Stevensa, Justin K.P. Leea, Mohsen Sadatsafavib, Geoffrey K. Blaira,⁎

aDivision of Pediatric General Surgery, University of British Columbia, Vancouver, British Columbia, CanadabCollaboration for Outcomes Research and Evaluation, Faculty of Pharmaceutical Sciences, University of British Columbia,Vancouver, British Columbia, Canada

Received 7 February 2009; revised 3 July 2009; accepted 3 July 2009

0d

Key words:Fine-needle aspiration;Thyroid nodules;Pediatric

AbstractBackground: There is conflicting evidence in the current literature regarding the use of fine-needleaspiration (FNA) biopsy in the diagnosis of a thyroid nodule in the pediatric population. There arenumerous studies that look at the sensitivity and specificity of this test with varying results. A meta-analysis will provide further insight into this topic.Purpose: Our objective was to investigate the diagnostic parameters of FNA biopsy in the diagnosis of athyroid nodule in the pediatric population, specifically, to determine the sensitivity, specificity, accuracy,positive predictive value (PPV), and negative predictive value (NPV) of the test in differentiatingmalignant vs benign tumors.Material and Methods: We performed a literature search of Medline, Embase, and evidence basedmedicine (EBM) reviews for English studies that looked at FNA biopsy in thyroid nodules in thepediatric population, in which diagnostic values were present or could be calculated and where FNAresults were compared to an acceptable reference standard. Two reviewers independently selected allabstracts, and from these, studies to review. Two reviewers also independently checked diagnosticvalues in the studies or calculated them from data available. A meta-analysis was performed, andpooled diagnostic test values were calculated using a random-effects, bivariate meta-regressionmodel. Studies were also assessed for quality using the quality assessment for diagnostic accuracystudies tool.Results: Twelve studies were included for review. The quality of the studies in general was good.The pooled estimate of sensitivity and specificity were 94% (95% confidence interval [CI], 86%-100%) and 81% (95% CI, 72%-91%), respectively. Assuming 20% of nodules are malignant, theaccuracy, PPV, and NPV were 83.6%, 55.3%, and 98.2%, respectively.Conclusion: This meta-analysis provides good evidence that FNA biopsy of thyroid nodules is asensitive test in the pediatric population and may be a useful tool for excluding malignancy inyoung patients. Future prospective studies are needed to evaluate this further.© 2009 Elsevier Inc. All rights reserved.

⁎ Corresponding author.E-mail address: gblair@cw.bc.ca (G.K. Blair).

022-3468/$ – see front matter © 2009 Elsevier Inc. All rights reserved.oi:10.1016/j.jpedsurg.2009.07.022

Thyroid nodules are relatively rare in the pediatricpopulation compared to the adult population, affecting1.6% to 1.8% of children and adolescents [1,2]. However,the risk of malignancy at presentation is much higher; 26.4%

2185Pediatric thyroid fine-needle aspiration cytology

in a recent meta-analysis [3] compared to a 5% risk ofmalignancy in adults [4]. In a recent multicenter study, theincidence of thyroid malignancy was found to be 43% inoperated patients [5]. Thus, the presentation of a patient witha thyroid nodule in pediatrics represents a diagnosticchallenge for physicians as management decisions dependon whether a nodule is malignant or benign. As part of theclinical workup for thyroid nodules, a variety of diagnosticstudies are indicated to detect or dismiss malignancy,including thyroid function tests (eg, serum thyroid hormonesT3 and T4, thyroid-stimulating hormone, and thyroidantibodies), radioiodine scanning to isolate hypofunctioningor hyperfunctioning nodules, ultrasound imaging, and fine-needle aspiration (FNA) biopsy [6,7]. One of the reasonsthyroid nodules represent a diagnostic challenge is thatphysicians have trouble interpreting the results of thediagnostic tests. For example, 2 of the aforementionedpreoperative tests, ultrasound and thyroid scintigraphy, havebeen shown to be poor at differentiating between benign andmalignant disease in the pediatric population [8-10]. Inaddition, despite the current evidence against the routine useof thyroid scintigraphy as a diagnostic tool to differentiatebenign from malignant nodules, in adult populations, 23%and 66% of North American and European endocrinologists,respectively, admitted to routinely ordering the test [11,12]—a practice that is mirrored in pediatric populations [8,9,13].

Of the available preoperative diagnostic tools, FNAbiopsy seems to be the most accurate tool in the diagnosisof a thyroid nodule. In adults, FNA has been established inlarge reviews as a useful test in the differentiation of athyroid nodule as malignant or benign [14,15] and in the1980s was endorsed by the American Thyroid Association asan accurate first-line test [16]. As a result, FNA hassignificantly reduced the number of surgeries for nodulespreoperatively diagnosed as benign.

Because of the higher risk of malignancy in pediatricthyroid nodules, it is common practice in many institutions toremove surgically any nodule by either a hemithyroidectomyor total thyroidectomy, with FNA playing a limited role inthe treatment algorithm. This is partly because of the lack ofconsensus regarding the use of FNA of thyroid nodules in thepediatric population. The literature on pediatric FNA issparse and often in disagreement with each other. Manyresearch groups have found that FNA in children andadolescents is a highly sensitive and accurate test indifferentiating benign from malignant nodules [8,9,17,18].Of these, most groups have also suggested that FNA canreduce the need for surgery for those with benign cytologicresults [8,9,17-19]. However, other studies have shown theopposite—that FNA in pediatrics is not as accurate asphysicians had hoped for and that surgery and subsequentpathologic diagnosis is still required to rule out a malignantthyroid nodule [20,21]. Furthermore, many clinicians feelthat FNA is not as practical in the pediatric populationcompared to the adult population because of patientdiscomfort, requirement of general anesthesia, or complica-

tions associated with the procedure. However, many researchgroups have found that the procedure is usually done withoutcomplication or significant patient discomfort [8,21] and canbe done without sedation or anesthesia [9,21].

The use of FNA for diagnosis of thyroid nodules in thepediatric population has been researched frequently in thelast 2 decades [8,9,13,17-29]. We performed a meta-analysisof the existing literature to determine the diagnostic value ofFNA in determining malignancy in children presenting witha thyroid nodule. Specifically, in the context of pediatricthyroid disease, we looked at the sensitivity, specificity,accuracy, and negative and positive predictive values (NPVand PPV) of FNA.

1. Methods

1.1. Search methods

Acomputerized database searchwas performed ofMedline,Embase, and EBM reviews (Cochrane Database of SystematicReviews, American College of Physicians (ACP) JournalClub, Database of Abstracts of Reviews of Effects, andCochrane Central Register of Controlled Trials). The searchincluded manual searches of bibliographies and any cross-references to the articles. Optimally sensitive search strategiesdeveloped byHaynes andWilczynski [30] andWilczynski et al[31] were used for Medline and Embase, respectively. Asimilar search strategy was also used for EBM reviews(Cochrane Database of Systematic Reviews, ACP JournalClub, Database of Abstracts of Reviews of Effects, andCochrane Central Register of Controlled Trials). These searchmethods included all studies from 1950 to December 2006.

1.2. Selection criteria

The following selection criteria were used:

1. The study investigated the diagnostic accuracy ofFNA biopsy in presenting thyroid nodules in apediatric population.

2. The study included diagnostic values (sensitivity andspecificity) or provided sufficient data that these valuescould be calculated.

3. The study used as a reference standard either (a)surgical pathologic specimen or (b) follow-up ofobserved nodules for a minimum 1 year for all patientsor mean minimum of 18 months.

4. The study population was children (age, b18-21) whopresented with undiagnosed thyroid nodules.

5. The studies were written in English.6. Case studies were not included.

Two reviewers independently read all abstracts. Articlesthat could not be excluded based on title or abstract were

Table 1 Items for quality assessment

Item 1 Was the spectrum of patients representative of thepatients who will receive the test in practice?

Item 2 Were selection criteria clearly described?Item 3 Is the reference standard likely to classify the target

condition correctly?Item 4 Is the period between reference standard and index test

short enough to be reasonably sure that the targetcondition did not change between the 2 tests?

Item 5 Did the whole sample or a random selection of thesample receive verification using a reference standarddiagnosis?

Item 6 Did patients receive the same reference standardregardless of the index test results?

Item 7 Was the reference standard independent of the indextest (ie, the index test did not form part of the referencestandard?)

Item 8 Was the execution of the index test described insufficient detail to permit replication of the test?

Item 9 Was the execution of the reference standard describedin sufficient detail to permit replication of the test?

Item 10 Were the index test results interpreted withoutknowledge of the results of the reference standard?

Item 11 Were the reference standard results interpreted withoutknowledge of the index test?

Item 12 Were the same clinical data available when the testresults were interpreted as would be available when thetest is used in practice?

Item 13 Were uninterpretable/intermediate results reported?Item 14 Were withdrawals from the study explained?

2186 C. Stevens et al.

retrieved in full text. These full text articles were read andincluded by 2 reviewers independently. If there wasdisagreement among reviewers regarding the inclusion ofan article, a third reviewer was consulted. Finally, an expertin the field was consulted to exclude studies for poor externalvalidity or poor methodological quality.

Table 2 Data for sensitivity, specificity, accuracy, NPV, and PPV

Term Cytologic examinatio

True-positive (TP)—traditional definition Malignant or suspici(including any follicu

True-negative (TN)—traditional definition BenignFalse-positive (FP)—traditional definition Malignant or suspici

(including any follicuFalse-negative (FN)—traditional definition BenignTrue-positive (TP)—traditional definition Malignant or suspici

(including any follicuTrue-negative (TN)—traditional definition BenignFalse-positive (FP)—traditional definition Malignant or suspici

(including any follicuFalse-negative (FN)—traditional definition Benign

Sensitivity = TP/(TP + FN). Specificity = TN/(TN + FP). PPV = TP/(TP + FP)

1.3. Quality

To assess the quality of the included articles, the “qualityassessment for diagnostic accuracy studies” (QUADAS) wasused [32]. The items for quality assessment are shown inTable 1. The quality of each study was independentlyevaluated by 2 reviewers using the QUADAS tool, and ascore for each item for each study was agreed upon.

1.4. Data extraction

Data for sensitivity, specificity, accuracy, NPV, and PPV(Table 2) were extracted from the included studies. Allreported diagnostic values were checked using raw datawhen provided in the study. If diagnostic values were notprovided and the data provided were sufficient, weindependently calculated values for sensitivity and specifi-city. This was done independently by 2 reviewers to ensurecalculations were correct. In one study, the values ofsensitivity and specificity that were calculated from theraw data differed from those presented in the results of thearticle. In this case, the difference was because of differingdefinitions of false-positive and false-negative from theremainder of studies in the analysis. Thus, the calculatedvalues were used [25].

1.5. Statistical analysis

The random-effects, bivariate meta-regression modelproposed by Reitsma et al [33] for the joint meta-analysisof sensitivity and specificity was used for the analysis. Weused the refinement of the Reitsma's method (using the“exact” binomial distribution instead of its normal approx-imation) proposed by Chu and Cole [34] to tackle the smallsample size and observed sensitivity/specificity of 100% insome of the underlying studies. Both the sensitivity andspecificity of FNA were adjusted for the traditional or

n Histopathologic examination

ouslar neoplasm)

Malignant

Benign (including follicular adenoma)ouslar neoplasm)

Benign (including follicular adenoma)

Malignantouslar neoplasm)

Malignant or follicular neoplasms

Benign (not including follicular neoplasms)ouslar neoplasm)

Benign (not including follicular neoplasms)

Malignant or follicular neoplasms

. NPV = TN/(TN + FN).

tem

8Item

9Item

10Item

11Item

12Item

13Item

14To

talscore

NY

YY

YY

11N

YU

YY

Y11

NY

UY

YY

10N

YU

YY

Y10

NY

UY

YY

10N

YU

YY

Y9

NY

UY

YY

11N

YU

YY

Y11

UY

UY

YY

10N

YU

YY

Y9

NY

UY

YY

10N

YU

YY

Y11

2187Pediatric thyroid fine-needle aspiration cytology

nontraditional definition of the index test. We calculated thepooled sensitivity, specificity, the receiver operating char-acteristics (ROC), its area under curve, and positive andnegative likelihood ratios. An ROC curve is a graphical plotof the sensitivity vs 1 − specificity (ie, true-positive vs false-positive rate) of the test as its discrimination threshold isvaried. The area under curve, which is the area under theROC curve, provides a general measure of the discriminatoryability of the test that, unlike sensitivity and specificity, is notaffected by variable discrimination thresholds that differenttest takers (eg, the pathologists who reads the FNA) mightuse. On the basis of the pooled estimates of sensitivity andspecificity, we also calculated the accuracy (proportion ofpatients correctly classified), PPV, and NPV of the test byassuming that 20% of nodules are malignant. For all pooledestimates, the point estimate and corresponding 95%confidence interval were calculated. Because sensitivityand specificity were meta-analyzed jointly, besides aconfidence interval, we could calculate a 95% “confidenceellipse” representing the likely range for the joint values ofsensitivity and specificity.

To determine the robustness of the pooled estimates ofsensitivity and specificity, we performed a jackknifesensitivity analysis by excluding each study one at a timeand pooled the remaining studies. All analysis wereperformed in SAS version 9.1.3 (SAS Institute, Cary, NC).

Table3

Resultsof

thequ

ality

analysisusingtheQUADAStool

Study

Item

1Item

2Item

3Item

4Item

5Item

6Item

7I

Al-Shaikhet

al[23]

NY

YY

YN

YY

Amikrachiet

al[17]

YY

YY

NY

YY

Ardaet

al[9]

YU

YY

YN

YY

Ardito

etal

[13]

NY

YY

YY

YN

Chang

etal

[22]

YU

YY

NY

YY

Corrias

etal

[8]

NU

YY

YY

YU

Degnanet

al[24]

YY

YY

YN

YY

Gharibet

al[15]

YY

YY

YN

YY

Khu

rana

etal

[19]

YU

YY

YN

YY

Lug

o-Vincenteet

al[36]

UU

YY

NY

YY

Raabet

al[35]

YN

YY

YN

YY

Willgerodt

etal

[21]

YY

YY

NY

YY

N=No,

Y=Yes,U

=Unk

nown.

2. Results

2.1. Search and selection

Our search strategy of Medline resulted in 707 articleswith an additional 98 from Embase. No additional articleswere found using EBM reviews. From these abstracts, 69 fulltext articles were retrieved. Finally, 12 studies that had allnecessary components were included in this review[8,9,13,15,17,19,21-24,35,36]. The main reasons for exclu-sion were incorrect study population, lack of sufficientreference standard, study used, results admixed with fine-needle biopsies of other tumors, insufficient follow-up, nonumbers or insufficient data given for calculation ofdiagnostic values, and case studies. One seemingly strongstudy provided diagnostic values that were disparate from thedata provided, and the data provided were not sufficient toindividually calculate diagnostic values [18].

2.2. Quality

The results of the quality analysis using the QUADAStool are shown in Table 3. All studies received a QUADASscore of 9 to 11. Items 5 and 6, which refer to the unbiaseduse of reference standard, in particular, were frequentlyscored as “no.”Many studies did not explain the execution ofthe fine-needle biopsies in sufficient detail to permit

Table 4 Results table

Study (reference) QUADAS Type of study Patient population No. ofpatients

Participantsin whomresultsare based

Follow-up Sensitivity Specificity Accuracy PPV NPV

Al-Shaikh et al [23] 11 Retrospective Children with FNA from1987-1999

41 37 Surgical pathology ormean, 23 mo

100 85.7 86.5 28.6 100

Amrikachi et al [17] 11 Retrospective Reports of All FNA from 4hospitals of children age 10-21

32 31 Surgical pathology orminimum 1 y; range,1-17 y

100 64.7 80.7 70 100

Arda et al [9] 10 Prospective Children presenting withthyroid nodule

46 44 Surgical pathology orminimum 1 y; range, 1-5 y

100 95 95.5 66.7 100

Ardito et al [13] 10 Retrospective Children with thyroidectomiesfrom 1987-99, age 9-20

50 50 All surgical pathology 100 57.1 70 50 100

Chang and Joo [22] 10 Retrospective Children with FNA, age 2-21 51 16 All surgical pathology 100 62.5 81.25 72.7 100Corrias et al [8] 9 Retrospective Children with thyroidectomies,

age 8-1742 42 All surgical pathology 95 86.4 90.5 86.4 95

Degnan et al [24] 11 Retrospective Children with FNA from1985-94, age 8-18

18 16 Surgical pathology or 2patients with cystsresolving on FNA

72.7 80 75 88.9 57.1

Gharib andGoellner [15]

11 Retrospective Children with FNA from1983-1995, age b 17

47 41 Surgical pathology orminimum 1 y; range, 1-5 y

90 96.8 95.1 90 96.8

Khurana et al [19] 10 Retrospective Children with FNA, age 9-20 57 57 Surgical pathology orminimum 1 y; range, 1-4 y

92.9 81.4 84.2 61.9 97.2

Lugo-Vincenteand Ortiz [36]

9 Retrospective Children with thyroidectomiesfrom 1985-95, age 9-18

24 15 All surgical pathology 75 81.8 80 60 90

Raab et al [35] 10 Retrospective Children with FNA in 4hospitals, age 1-18 y

66 63 Surgical pathology orminimum 1 y; range, 1-5 y

88.9 92.6 92.1 66.7 98

Willgerodt et al [21] 11 Retrospective Children with FNA from1971-2001

169 63 All surgical pathology 63.6 78.9 77.2 26.9 94.7

2188C.

Stevenset

al.

Fig. 1 95% confidence ellipse and ROC of FNA. Continuous lineindicates ROC; dashed line, 95% confidence ellipse of the pooledestimates of sensitivity and specificity; crosses, sensitivity/specifi-city pairs of underlying studies.

2189Pediatric thyroid fine-needle aspiration cytology

replication of the test (item 8), and only one study explainedthe execution of the surgical pathologic examination as areference standard (item 9). Furthermore, in no study was itexplained whether the pathologist had knowledge of theresults of the fine-needle biopsy at the time of the referencestandard (item 11).

2.3. Results of diagnostic value of FNA

The obtained or calculated results of each study as well ascharacteristics of each study are shown in Table 4.

The 12 studies included in the final analysis havecollectively included 183 cases of malignant nodules and347 cases of benign nodules. Based on the traditionaldefinition of the gold standard, the pooled estimate ofsensitivity and specificity were 94% (95% confidenceinterval [CI], 86%-100%) and 81% (95% CI, 72%-91%). Ifthe gold standard is based on the nontraditional definition ofmalignancy, the pooled sensitivity and specificity were 82%(95% CI, 55%-100%) and 91% (95% CI, 18%-100%),respectively. Assuming 20% of nodules are malignant, theaccuracy, PPV, and NPV were 83.6%, 55.3%, and 98.2%,respectively. The jackknife sensitivity analysis indicated thatthe results were robust as the indices of test accuracy did notchange remarkably with the exclusion of individual studies.

The ROC curve of the test, along with the Sn/Sp pairs ofthe underlying studies and the 95% confidence ellipse for thepooled Sn/Sp is presented in Fig. 1.

3. Discussion

The purpose of this meta-analysis was to determine the testparameters for FNA biopsy as a preoperative diagnostic toolin presenting thyroid nodules in the pediatric population. Thismeta-analysis provides strong evidence that FNA biopsy is asensitive test for the detection of malignancy in thyroidnodules and should play a central role in determining which

patients should proceed to surgery and which patients shouldbe followed in clinic. We used a vigorous method for jointmeta-analysis of sensitivity and specificity, which unlike theunivariate meta-analysis, preserves the negative correlationthat often exists between sensitivity and specificity and henceprovides a robust pooled estimate [34]. The jackknifesensitivity analysis also indicates that the results are notsensitive to the inclusion of any particular individual study.

Using the traditional definition of false-positives andfalse-negatives, our analysis showed that FNA is a verysensitive test for the detection of malignancy. Furthermore,FNA has a high NPV, which represents the ability of the testto accurately predict benign nodules. This implies, clinically,that FNA is a useful test in ruling out malignancy in a benignthyroid nodule. This has great significance as it allowsphysicians to better predict which patients can avoid athyroidectomy for benign disease. The use of FNA and itsrole in ruling out malignancy has long been known andexploited in the adult population being widely accepted asthe “gold standard” for preoperative diagnosis [6]. Belfioreand La Rosa [37] suggest that FNA decreases the overall costof care by 25%, a saving of $500 to $1300 per patient, mostlyby avoiding unnecessary surgeries. These values wouldlikely be similar in the pediatric population.

Although our analysis showed that FNA is a sensitive testwith 94% sensitivity, it is not a specific one with a specificityof 81%. The reason for the tests' low specificity of 81% islargely explained by the precise definition of false-positivesand false-negatives. The sensitivity and specificity of theFNA is dependent on and varies with these definitions offalse-positive and false-negative. For instance, most studiesin our review chose to use what we coined the traditionaldefinition of false-positives and false-negatives. In thismodel, malignant or suspicious cytologic diagnoses that hada malignant lesion on surgical pathologic examination areconsidered true-positives. Follicular neoplasm seen oncytologic examination was considered part of the suspiciouscategory. Thus, follicular neoplasm on FNA with apathologic result of follicular adenoma would be considereda false-positive. In contrast, a few studies considered a true-positive as malignant or suspicious cytologic condition witha pathologic result of malignancy or follicular neoplasms,including follicular carcinoma or adenoma. When thisnontraditional definition of defining true-positives andtrue-negatives is used, specificity is increased to 91% andsensitivity is decreased from 94% to 82% compared to thetraditional definition of false-positives. Thus, using thetraditional definition maximizes sensitivity and NPV to thedetriment of specificity and PPV. Furthermore, the tradi-tional definition is more meaningful in the clinical setting asthe purpose of FNA as a diagnostic test is to determinewhich nodules are benign or malignant, not to correctlyidentify follicular neoplasms. In the above example,“suspicious” follicular neoplasms on cytologic examinationshould be treated as malignancy and at the present timearguably warrant at least a hemithyroidectomy. A histologic

2190 C. Stevens et al.

finding of benign neoplasm should be considered a false-positive as surgery was not, in hindsight, needed. If, after ahemithyroidectomy, the lesion proves histologically to be afollicular carcinoma, then a completion thyroidectomy maybe performed.

It is reasonable to suggest that the data in our meta-analysis is underestimating the potential of FNA in today'sclinical situation. This is for 2 reasons. One, this analysisincluded many studies that did not use ultrasound in theirFNA biopsies of the thyroid. At the institution of theseauthors, it is now standard that every biopsy is done withultrasound guidance. It is feasible that with ultrasoundguidance, the sensitivity and specificity of FNA biopsycould be slightly better. Second, it is apparent that there isinconsistency in the results that are reported by patholo-gists. For example, in this analysis, studies had categorizedFNA results as benign, suspicious, or malignant. However,in clinical situations, pathologist reports are often not sowell categorized or helpful. Many pathologists will giveobservational information without categorizing thesenodules in categories that can be correlated with thisstudy. Thus, FNA biopsy is only as good as the operatordoing the biopsy and whether they use ultrasound and thepathologist reading the report.

The main diagnostic pitfall of FNA is what to do with thefinding of follicular cells on FNA cytology. Follicular cellscan represent a malignant follicular carcinoma or a benignfollicular adenoma. Unfortunately, unlike papillary carci-noma, whether a follicular lesion is malignant or benigncannot be determined with FNA biopsy because tissuearchitecture and histologic examination is required. Inaddition, a high percentage of follicular cells representmalignancy. Thus, it may be argued that in a clinical setting,any follicular neoplasm should be treated as “suspicious formalignancy” and at a minimum, a hemithyroidectomywarranted. In some studies, a lower than expected valuefor sensitivity and specificity were obtained because of highrates of follicular neoplasms [4].

Although this analysis provides useful informationregarding the test parameters of FNA biopsy, there aredefinite limitations to this meta-analysis. One limitation isthat this analysis does not address unique populations ofpatients who present in the clinic, for example, a patient witha nonpalpable nodule, presumably detected on incidentalimaging. Few studies in this meta-analysis addressed theclinical presentation of the patients involved in their study. In3 studies, the presentation was mentioned [8,13,36], but onlyone study stated that they included patients with nonpalpablethyroid nodules [8]. Thus, it is still uncertain if nonpalpablenodules deserve the same treatment or have the same successwith FNA as palpable nodules. A patient with multiplethyroid nodules vs a solitary thyroid nodule is another factorthat is not specifically studied in this analysis. Six of thestudies stated that they included patients with multiplenodules [8,19,21,24,35], but it is uncertain as to whether thisfactor is important in diagnosis or outcome of these patients.

Finally, the importance of size of a thyroid nodule andspecifically what to do with micronodules (b1 cm) iscontroversial in today's literature [38]. Common sensesuggests that larger nodules would receive more aggressivetreatment of worry of malignancy. Unfortunately, it isbeyond this analysis to understand the significance ofmicronodules. Only 2 studies mentioned the size of thenodules in their study and neither included subset analysesfor this characteristic [8,9].

In the adult population, FNA is considered the mostaccurate test for preoperative diagnosis of thyroid nodulesand other tests such as ultrasound and thyroid scintigraphyhave been found to be quite unreliable [36]. Furthermore,there is sufficient evidence to suggest that these tests areunreliable for preoperative diagnosis in the pediatricpopulation as well [8,9]. Although we are not suggestingthat FNA should be used in isolation, FNA should play acentral role in pediatric practice in the algorithm of thediagnosis of a thyroid nodule and, in the right clinicalsetting, may prove with future needed research at hand tobe a useful screening tool in excluding malignancy. Ofcourse, FNA is only one tool of many to help diagnosethyroid nodules and decide further treatment. Clinicalparameters such as medical history and exposure toprevious radiation, family history for medullary thyroidcarcinoma, size, and mobility of the nodule, and noduletenderness, should still be considered when deciding furthertreatment or labeling a nodule as probable benign ormalignant. Furthermore, given the fallibility of anydiagnostic test, it is still reasonable to proceed directly tosurgery in cases where a nodule is very likely malignant,such as patients with significant radiation exposure to theneck or family history of medullary thyroid cancer.

There is good evidence that FNA in the pediatricpopulation is a sensitive diagnostic test and is a useful toolfor excluding malignancy. There is limited research in thisarea, and large studies are rare. In the meantime, cautionmust be observed in placing too much faith on the NPV ofthyroid FNA testing in children and youth. Furtherresearch is needed in the form of prospective studies tolook at the accuracy of FNA and its use in the algorithm ofthe management of patients presenting with thyroidnodules in pediatrics.

References

[1] Rallison ML, Dobyns BM, Keating Jr FR, et al. Thyroid nodularity inchildren. JAMA 1975;233(10):1069-72.

[2] Rallison ML, Dobyns BM, Meikle AW, et al. Natural history of thyroidabnormalities: prevalence, incidence, and regression of thyroiddiseases in adolescents and young adults. Am J Med 1991;91(4):363-70.

[3] Niedzela M. Pathogenesis, diagnosis and management of thyroidnodules in children. Endocr Relat Cancer 2006;13:427-53.

[4] Yeung MJ, Serpell JW. Management of the solitary thyroid nodule.Oncologist 2008;13:105-12.

2191Pediatric thyroid fine-needle aspiration cytology

[5] The Canadian Pediatric Thyroid Nodule Study Group. The CanadianPediatric Thyroid Nodule Study: an evaluation of current managementpractices. J Pediatric Surg 2008;43:826-30.

[6] Kim N, Lavertu P. Evaluation of a thyroid nodule. Otolaryngol ClinNorth Am 2003;36(1):17-33.

[7] Daneman D, Daneman A. Diagnostic imaging of the thyroid andadrenal glands in childhood. Endocrinol Metab Clin North Am2005;34(3):745-68, xi.

[8] Corrias A, Einaudi S, Chiorboli E, Weber G, Crino A, Andreo M, et al.Accuracy of fine needle aspiration biopsy of thyroid nodules indetecting malignancy in childhood: comparison with conventionalclinical, laboratory, and imaging approaches. J Clin Endocrinol Metab2001;86(10):4644-8.

[9] Arda IS, Yildirim S, Demirhan B, et al. Fine needle aspiration biopsyof thyroid nodules. Arch Dis Child 2001;85(4):313-7.

[10] Lawrence Jr W, Kaplan BJ. Diagnosis and management of patientswith thyroid nodules. J Surg Oncol 2002;80(3):157-70.

[11] Bennedbaek FN, Perrild H, Hegedus L. Diagnosis and treatment of thesolitary thyroid nodule. Results of a European survey. Clin Endocrinol(Oxf) 1999;50(3):357-63.

[12] Bennedbaek FN, Hegedus L. Management of the solitary thyroidnodule: results of a North American survey. J Clin Endocrinol Metab2000;85(7):2493-8.

[13] Ardito G, et al. Thyroid tumors in children and adolescents:preoperative study. Eur J Pediatric Surg 2001;11:154-7.

[14] Ashcraft MW, Van Herle AJ. Management of thyroid nodules. II:scanning techniques, thyroid suppressive therapy, and fine needleaspiration. Head Neck Surg 1981;3(4):297-322.

[15] Gharib H, Goellner JR. Fine-needle aspiration biopsy of the thyroid: anappraisal. Ann Intern Med 1993;118(4):282-9.

[16] Kaplan MM. Evaluation of a thyroid nodule by needle aspiration. In:Braverman LE, Utiger RD, editors. The thyroid. 8th ed. Philadelphia:Lippincott Williams & Williams; 2000. p. 441-51.

[17] Amrikachi M, Ponder TB, Wheeler TM, et al. Thyroid fine-needleaspiration biopsy in children and adolescents: experience with 218aspirates. Diagn Cytopathol 2005;32(4):189-92.

[18] Hosler GA, Clark I, Zakowski MF, et al. Cytopathologic analysis ofthyroid lesions in the pediatric population. Diagn Cytopathol 2006;34(2):101-5.

[19] Khurana KK, Labrador E, Izquierdo R, et al. The role of fine-needleaspiration biopsy in the management of thyroid nodules in children,adolescents, and young adults: a multi-institutional study. Thyroid1999;9(4):383-6.

[20] Lugo-Vicente H, Ortiz VN, Irizarry H, et al. Pediatric thyroid nodules:management in the era of fine needle aspiration. J Pediatr Surg 1998;33(8):1302-5.

[21] Willgerodt H, Keller E, Bennek J, et al. Diagnostic value of fine-needleaspiration biopsy of thyroid nodules in children and adolescents. JPediatr Endocrinol Metab 2006;19(4):507-15.

[22] Chang SH, Joo M, Kim H. Fine needle aspiration biopsy of thyroidnodules in children and adolescents. J Korean Med Sci 2006;21(3):469-73.

[23] Al-Shaikh, et al. Fine-needle aspiration biopsy in the management ofthyroid nodules in children and adolescents. J Pediatr 2001;138:140-2.

[24] Degnan BM, et al. An analysis of fine-needle aspiration biopsy of thethyroid in children and adolescents. J Pediatr Surg 1996;31(7):903-7.

[25] Gharib H, Zimmerman D, Goellner JR, et al. Fine-needle aspirationbiopsy: use in diagnosis and management of pediatric thyroid diseases.Endocr Pract 1995;1(1):9-13.

[26] Eisenhut CC, et al. Fine-needle biopsy of pediatric lesions: a three-yearstudy in an outpatient biopsy clinic. DiagnCytopathol 1996;14(1):43-50.

[27] Mobley DL. Fine-needle aspiration biopsy: application to pediatrichead and neck masses. Laryngoscope 1991;101:469-72.

[28] Millman B, Pellitteri PK. Nodular thyroid disease in children andadolescents. Otolaryngol Head Neck Surg 1996;116:604-9.

[29] Silverman JF. Pediatric fine-needle aspiration biopsy. Am J Clin Pathol1991;95:653-9.

[30] Haynes RB, Wilczynski NL. Optimal search strategies for retrievingscientifically strong studies of diagnosis from Medline: analyticalsurvey BMJ, doi:10.1136/bmj.38068.557998.EE.

[31] Wilczynski A, et al. EMBASE search strategies for identifyingmethodologically sound diagnostic studies for use by clinician andresearchers. BMC Med 2005;3(7).

[32] Whiting P, Rutjes A, Reitsma J, et al. The development of QUADAS: atool for the quality assessment of studies of diagnostic accuracyincluded in systematic reviews. BMC Med Res Methodol 2003;3:25.

[33] Reitsma JB, Glas AS, Rutjes AW, et al. Bivariate analysis of sensitivityand specificity produces informative summary measures in diagnosticreviews. J Clin Epidemiol 2005;58(10):982-90.

[34] Chu H, Cole SR. Bivariate meta-analysis of sensitivity and specificitywith sparse data: a generalized linear mixed model approach. J ClinEpidemiol 2006;59(12):1331-2.

[35] Raab SS, Silverman JF, Elsheikh TM, et al. Pediatric thyroid nodules:disease demographics and clinical management as determined by fineneedle aspiration biopsy. Pediatrics 1995;95(1):46-9.

[36] Lugo-Vicente H, Ortiz VN. Pediatric thyroid nodules: insights inmanagement. Bol Asoc Med P R 1998;90(4-6):74-8.

[37] Belfiore A, La Rosa GL. Fine-needle aspiration biopsy of the thyroid.Endocrine Metab Clin North Am 2001;30:361-400.

[38] Butros R, Boyvat F, Ozyer U, et al. Management of infracentimetricthyroid nodules with respect to ultrasonographic features. Eur Radiol2007;17(5):1358-64.