Objective: The usefulness of repeated fine-needle cytology (FNC) in thyroid nodules with benign cytology remains unknown. We analyzed the relevance of repeated FNC to detect suspicious or malignant (S/M) cytologies and carcinomas.
Design: A retrospective study (1983–2004) was conducted in our endocrinology department.
Methods: We reviewed the reports of 895 adequate FNC performed in 298 patients (298 nodules) during a mean follow-up of 5 years. We compared the nodules with at least one suspicious or malignant FNC (S/M nodules) with nodules with repeatedly benign (RB) FNC (RB nodules).
Results: Among the nodules with initial benign cytology, we found 35 nodules with one or more later suspicious or malignant results. The interval between the first FNC and the first S/M FNC was 2.9 years. The probability for a nodule to have a repeated benign FNC decreases with time and with the number of FNC. We did not find any clinical or ultrasonographic characteristics related to an S/M cytology. Seven cancers were detected by the second or the third FNC with S/M results. The proportion of cancers among S/M nodules was similar when S/M cytology appears during the first, the second, or the third FNC.
Conclusions: We suggest to repeat FNC up to three adequate samples in the follow-up of thyroid nodules so as not to miss the presence of malignant neoplasm.
Thyroid nodules are frequent since prevalence is estimated as 4–7% of the adult population and annual incidence 100 per 100 000 (1). The majority of these nodules are benign tumors and cancers represent 3–10% of cases (2). The main objective in the management of thyroid nodules is to differentiate cancers from benign neoplasms. To that aim, various clinical criteria were previously proposed such as male gender, age <20 or >60 years (3), hard consistency, spherical shape (4), previous childhood irradiation (5), or some family history such as Gardner’s syndrome (6). Some ultra-sonographic features are also suggestive of malignancy, such as the absence of hypoechogenic halo surrounding the lesion and irregular margin (7), microcalcifications, or marked intranodular blood flow (8). However, the specificity and sensitivity of these characteristics are not high enough to help in making a decision (1).
Fine-needle cytology (FNC) is widely accepted (9, 10) as a first-line procedure for selection of nodules to be operated. The results are usually put into four groups: benign, suspicious, malignant, and inadequate. Surgical excision is recommended for all malignant cytologies (11) and also advocated for suspicious cytologies. The latter mainly correspond to nodules with cell anomalies and to follicular or Hürthle cell neoplasms in which cancers may represent up to 29% (12). Inadequate cytologies very often correspond to samples with a low cell content, vascular lesions, or necrotic changes. Twenty percent of cancers are in the form of cystic lesions, and cystic changes appear in 23–33 and 27–35% of benign and malignant nodules respectively (13). Repeated FNC is thus recommended when the samples are inadequate. However, there is no consensus for the need to repeat cytology after an initial benign result, despite the reported minimal 1% rate of false-negative results (14). Some authors advise performing reaspira-tion if the nodule grows (15) and others repeat FNC only once, one year after the initially benign FNC or in case of a change in the consistence of the nodule (16). Although these attitudes seem appropriate considering the 65–98% FNC sensitivity (17), few studies have assessed the efficacy of repeated cytology in apparently benign nodules.
Subjects and methods
We retrospectively revised the reports of cytology for patients who repeatedly attended our department between January 1983 and June 2004. We perform 800 FNC per year (60% benign, 17% suspicious, 5% malignant, and 18% inadequate results (18)). Inclusion criteria were: presence of at least two consecutive adequate FNC of the same palpable nodule, be it solitary or within multinodular goiter, at least one ultrasound with clearly defined limits of the nodule, and the same pathologist for analysis of the FNC. The retrospective nature of the study and the composite inclusion criteria lead us to restrict our analysis to 298 nodules (298 patients, 895 FNC).
Only palpable nodules were studied to ensure proper location of the needle, since the FNC were performed without ultrasound guidance. The puncture was always performed by the referent physician, as previously described (19), with a 23-gauge needle without aspiration except for cystic lesions. The smears were spread on slides and air-dried before May–Grunwald–Giemsa staining. In the case of multinodular goiter, only the largest nodule was studied. According to the guidelines of the Papanicolaou Society of Cytopathology (20), smears were considered adequate if they contained at least five groups of 10–15 cells each. Results were sorted into three categories: (1) benign lesions including adenomatous or colloid nodule eventually with cystic changes and Hashimoto’s thyroiditis, (2) suspicious lesions which include indeterminate lesions (follicular neoplasms, oxyphilic tumor) and lesions with limited features suggestive of but not definitive for malignancy, and (3) malignant lesions. Inadequate cytologies (insufficient number of cells) were excluded.
All cytologies were analyzed by only one trained cytopathologist aware of the prior FNC results. In 2006, all suspicious and malignant FNC were reanalyzed by a second cytopathologist to make sure all the presently accepted criteria of suspicious cytology were present.
Thyroid ultrasonography was performed to assess the number, size and echostructure of the nodule, looking for suspicious features: microcalcifications, absence of a peripheral hypoechogenic halo, or irregular margins. For 216 patients, ultrasonography was repeated at the end of the follow-up, allowing for the calculation of a percentage of growth: ((initial diameter(final diameter)/initial diameter))×100. Serum TSH, serum-free T4 (FT4), thyroid antibodies (peroxidase (TPOab) or thyroglobulin antibodies (Tgab)) and calcitonin were measured. Radionuclide (I123 or Tc99m) scanning was also performed.
Any suspicious or malignant cytology as well as progression in size, appearance of a hard consistence of the nodule, or presence of adenopathy led to surgery. Surgery was also decided in case of ultrasonographic suspicious features as described above. A total of 16 patients with a first suspicious FNC initially refused surgery and preferred having a second FNC.
The endpoint of the study was the appearance of a suspicious or malignant cytology. Suspicious/malignant (S/M) nodules correspond to all the nodules having at least one suspicious or malignant cytology as opposed to repeatedly benign (RB) nodules which always had benign cytology. For each nodule, only results of adequate FNC were analyzed. Only the first S/M cytology was considered for calculation of the time and the number of FNC between the baseline FNC and the endpoint. Microcarcinomas were excluded from the analysis.
The Student’s t-test and Mann–Whitney test, when appropriate, were used to compare continuous variables. The χ2 test or Fisher’s exact test, when appropriate, was used for comparison of categorical variables. The probability of an adequate benign cytology during the follow-up of patients was calculated using Kaplan–Meier techniques. All P values were two-sided and a P value <0.05 indicated statistical significance. All analyses were performed by Systat statistical software (version 9.0; SSI, San Jose, CA, USA).
Table 1 summarizes the characteristics of patients and nodules. At baseline, patients (95% female, 5% male) had a mean age of 45.7 years. According to palpation, nodules were solitary in 70.6%, but on ultrasound they appeared to be part of a mutinodular goiter in 42.1% of cases. Ultrasonographic diameter was 19.5 mm (±9.3). On radionuclide scanning, 96% of nodules appeared non-functioning. Most patients (92.1%) were euthyroid, 5.5% were hyperthyroid, and 2.4% were hypothyroid.
The FNC results are shown in Fig. 1 and Table 2. We performed 895 FNC, a mean of 3.0 FNC per nodule (range 2–5 FNC), and three or more FNC were recorded in more than 50% of nodules. The mean interval between the first and the last FNC was 5.0 years (range 1–192 months). The initial cytology was benign in 94.6% (282), suspicious in 5.4% (16), and malignant in 0% of nodules. Out of 298 nodules, 82.9% (247) had a repeatedly benign cytology and 17.1% had at least one suspicious (50) or malignant FNC (1). Out of these 51 S/M nodules, 35 were suspicious or malignant at the time of subsequent FNC and 16 were initially suspicious, including 10 for which the S/M characteristic was confirmed on repeated FNC. A total of 41 nodules were only once suspicious or malignant and 10 were several times. The mean interval between first FNC and first S/M FNC was 2.9 years (range 0–11 years). Starting from 5.4% of first FNC in the nodules studied here (surgery refused despite an S/M FNC), the frequency of S/M FNC reached 11.1, 14.8, 16.8, and 17.1% after the second, third, fourth, and fifth FNC respectively. Eighty-six percent of S/M FNC were found at one of the three first cytologies. Whereas only 5.7% of the nodules were described as suspicious lesions after 6 months, they were 6, 10.1, and 14.4% after 1, 3, and 5 years respectively.
In 298 patients, 119 (39.9%) were operated on average 4.1 years (range 1–198 months) after the first FNC and generally after one of the three first FNC. Results are shown in Tables 2 and 3. They split in 86% of S/M nodules (44 out of 51) and 30% of RB nodules (75 out of 247). Among these 119 operated nodules, 11 (9.4%) were malignant and corresponded to 10 papillary and 1 follicular carcinomas (Fig. 1). However, 22.7% of S/M nodules corresponded to cancers and conversely 91% (10 out of 11) of the malignant nodules had had at least one S/M cytology. All but one malignant neoplasms were identified by an S/M FNC performed 3.4 years (range 0–9.8 years) after the first FNC. Three carcinomas were found among the 13 initially S/M operated nodules (Tables 3 and 4). The seven other cancers were detected by the second or the third FNC with S/M results (only 31 out of 35 S/M nodules were operated). Among these seven carcinomas, one corresponded to stage II and one to stage III on new stage grouping of TNM classification for thyroid cancer (21); all the others corresponded to stage I. Five cancers were found, thanks to a subsequent cytology with suspicious results, one with malignant result and one with indeterminate result (Table 4). The proportion of cancers among operated S/M nodules was not significantly different when the S/M cytology appeared at the first (23.1%), the second (19%), or the third (44%) FNC. The fourth or fifth FNC with S/M results never corresponded to carcinomas (Table 3). There was no difference in the risk of malignancy between the nodules with only one or more S/M results (20% versus 33.3%). Out of the 75 operated RB nodules, because of modifications of clinical characteristics, only one (1.3%) was malignant and had had only two FNC.
These data allowed evaluation of the performance of the repetition of FNC (Table 4). The number of false-negative results decreases from eight (6.7%) at the first FNC to one (0.8%) after third FNC and, conversely, the number of false-positive results increases from 10 to 34.
We compared the characteristics of S/M and RB nodules (Table 1) and did not find any significant difference in age, sex, or thyroid function tests between patients with RB or S/M nodules. All ultrasonographic characteristics (largest diameter, diameter variation and proportion of isolated or multiple nodules) were also similar. The frequency of suspicious ultrasonographic features was not significantly different between S/M and RB nodules, nor between proven cancers and benign lesions in operated nodules. The results of radionuclide scanning were also identical.
We had all 51 S/M FNC reviewed by a different cytopathologist. All results but one were confirmed. One lesion was initially described as a follicular adenoma with suspicious oxyphilic cells and subsequently as a benign lesion with some oxyphilic cells.
Malignant lesions are the main indications for surgery, but most authors also agree to propose surgery for nodules with suspicious results (1, 10, 11). A single FNC is proposed in the management of thyroid nodules according to the high sensitivity and specificity of the cytology observed by trained groups depending on whether suspicious and inadequate cytologies are taken into account (17). Nevertheless, several authors studied the usefulness of repeating FNC in apparently benign nodules (22–27). This could decrease the false-negative results from 5.2% to less than 1.3% (25). However, some series concluded a limited benefit of repeating FNC (28–30). This retrospective study concerns 298 palpable nodules analyzed by the same trained pathologist, excluding all inadequate specimens. Thus, in a 5-year long follow-up, we could analyze the results of 895 adequate FNC with two FNC for all and three or more FNC for half of the nodules. Therefore, to our knowledge, our study is the longest and one of the largest published to date. We detected 35 additional S/M nodules including seven cancers. Also, the false-negative results decreased from 6.7% at the first adequate FNC to 0.8% after the repeated FNC. Moreover, two out of these seven cancers corresponded to stage II or III and required a more intensive therapy. These cancers corresponded equally to S/M nodules with suspicious, malignant, or only indeterminate results.
A crucial question addressed by the aforementioned studies is the minimal number of FNC necessary for accuracy of the benign nature of a nodule and the characteristics predicting the appearance of a suspicious cytology. In our study, all but one carcinomas were found by an S/M cytology out of one of the first three FNC and all S/M cytologies beyond the third FNC corresponded to benign lesions. Only one operated RB nodule corresponded to a papillary carcinoma but only two FNC had been performed on this nodule. It has been proposed that a second cytology is enough providing the nodule remains clinically unchanged in size and consistency (15, 16, 27). We did not find any significant difference in initial size, growth, consistency or number of nodules between the repeatedly benign and S/M nodules. The non-functioning feature of a nodule, proposed to select suspicious nodules that would imply surgery (15), did not help here. Therefore, the repetition of FNC, up to three adequate specimens, seems to be more efficient than any other procedure to limit the risk of missing a diagnosis of cancer.
The duration of follow-up has also been debated. A 6–12 month interval between the first and second FNC has been proposed (31), but data from longer studies are missing in the literature. We showed that initially benign nodules can become suspicious even after several years. Seven carcinomas were found, thanks to an S/M cytology done on average 3 years after the first benign one.
Some limitations are present in this study. First, because it is retrospective, the number of FNC, the delay between two FNC and the length of the follow-up are not the same for all the nodules. However, this may have underestimated the actual number of S/M cytologies or cancers. Second, the endpoint we used to evaluate the usefulness of repeating FNC was an S/M cytology not a proven cancer which could appear a more relevant criteria to assess a strategy, though biased as long as most patients are not operated. It is noteworthy that we have found that 22.7% of all operated S/M nodules were cancers in keeping with the literature (15, 32) in contrast with 1.3% cancers in operated RB nodules. The frequency of confirmed cancer did not significantly change according to the rank of the S/M FNC, although small differences in the performance of the different FNC may have been missed owing to the small number of cases. Therefore, despite the increase in ‘undue’ surgical procedures decided on the basis of an S/M for nodules initially considered benign, any nodule with suspicious or malignant cytology even after one or several benign FNC implies surgery. Some authors have found that immunocytochemistry (galectin-3, human bone marrow endothelial cell-1, cytokeratin-19) may be useful in the preoperative evaluation of thyroid nodules (33, 34). Although immunocytochemistry can hardly be used on a routine basis, it could be advised on a repeated FNC after a first S/M, to reduce the number of false-positive results.
The increase in the S/M FNC relates to several explanations. Obviously, one is the increase in false-positive results. More interesting is the observed decrease in false-negative results. Several biases have been avoided. The same trained senior pathologist analyzed the successive FNC, thus excluding the interindividual variability and the intraindividual improvement observed in novice pathologists. Ultrasound-guided FNC were not included here because it was not routinely used at the start of the follow-up. This limits the study to unambiguously palpable nodules and rules out a decrease in false-negative results solely due to an improvement in sampling accuracy (35). However, since only adequate cytologies were recorded, our results would likely not have been changed by ultrasound guidance. Finally, the increase in S/M FNC and subsequently in proven cancers is not due to an increase in frequency of microcarcinomas as previously reported (36), since they were excluded of the analysis.
A cost-effectiveness analysis of repeating FNC, not yet available though mandatory, should consider not only additional costs (of FNC, immunocytochemistry, undue surgery) but also the putative benefit of an early diagnosis of a thyroid cancer when compared with a later diagnosis.
In conclusion, the repetition of FNC reveals a relevant number of cancers which were missed by the first FNC. According to our results, a minimum of three adequate FNC seems appropriate to decrease the number of false-negative results. Any other criterion during the follow-up appears of less value than repeating FNC and may lead to an increase in unnecessary surgery. The determination of the optimal interval between the different FNC and duration of follow-up would require a larger series with a longer follow-up.
Baseline characteristics of 298 patients and nodules according to the existence of suspicious or malignant cytology during follow-up.
|All nodules (n=298)||RB nodules (n=247)||S/M nodules (n=51)||P*|
|F, female; M, male. RB nodules, nodules with repeatedly benign cytologies; S/M nodules, nodules with at least one suspicious or malignant cytology.|
|*RB nodules versus S/M nodules. P<0.05 was considered as significant.|
|Sex (F/M) (%)||94.9/5.1||95.9/4.1||90.2/9.8||0.15|
|Age (year)||45.7 (±15.5)||45.6 (±15.6)||46.3 (±15.3)||0.78|
|Clinical characteristics (%)|
|Solitary nodule/multinodular goiter||70.6/29.4||70.6/29.4||70.6/29.4||0.99|
|Suspicious findings (no/yes)||90.5/9.5||91.0/9.0||88.2/11.8||0.60|
|Diameter (mm)||19.2 (±10.1)||19.1 (±9.6)||19.8 (±12.3)||0.71|
|Ultrasonographic characteristics (%)|
|Solitary nodule/multinodular goiter||57.8/42.2||58.4/41.6||55.1/44.9||0.67|
|Suspicious features (no/yes)||92.1/7.9||91.3/8.7||96.0/4.0||0.39|
|Diameter (mm)||19.5 (±9.3)||19.3 (±8.8)||20.4 (±11.5)||0.47|
|Diameter variation (%)||+10.5 (±0.5)||+10.8 (±0.5)||+9.1 (±0.5)||0.87|
|Thyroid antibodies (%)||17.9||17.7||19.5||0.82|
Results of all 895 adequate FNC.
|FNC, fine-needle cytology. All inadequate cytologies were excluded (20.8% at first, 17.8% at second, 15.6% at third, 14.4% at fourth, and 6% at fifth).|
|Number (% of nodule)||298 (100)||298 (100)||181 (60.7)||92 (30.9)||27 (9.1)||298 (100)||895|
|Benign (%)||282 (94.6)||273 (91.6)||164 (90.6)||85 (92.4)||26 (96.3)||258 (86.6)||830 (92.7)|
|Suspicious (%)||16 (5.4)||25 (8.4)||16 (8.8)||7 (7.6)||1 (4.7)||39 (13.1)||64 (7.2)|
|Malignant (%)||0||0||1 (0.6)||0||0||1 (0.3)||1 (0.1)|
Number of proven cancers in the operated nodules with successive S/M FNC.
|FNC, fine-needle cytology; S/M cytology, cytology with suspicious or malignant result. Please note that only 44 out of 51 nodules with S/M FNC were operated.|
|First S/M cytology||13||16||9||5||1||44|
|Number of cancer (%)||3 (23.1)||3 (18.8)||4 (44.4)||0 (0)||0 (0)||10 (22.7)|
Performance of initial and repeated adequate fine-needle cytologies in 119 operated nodules.
|Benign (n=108)||Malignant (n=11)||False- negative resultsn (%)||False- positive resultsn (%)|
|FNC, fine-needle cytology; S/M, suspicious or malignant.|
|Initial FNC||8 (6.7)||10 (8.4)|
|Repeated FNC||1 (0.8)||34 (28.6)|
RagoT Vitti P Chiovato L Mazzeo S De Liperi A Miccoli P Viacava P Bogazzi F Martino E & Pinchera A. Role of conventional ultrasonography and color flow–doppler sonography in predicting malignancy in ‘cold’ thyroid nodules. European Journal of Endo-crinology199813841–46.
ShahJP Kian K Forastiere A Garden A Hoffman HT Jack Lee J Lydiatt W Medina JE Mukherji S Oliva ME O’Sullivan B Paulino A Singh B Weber R & Weymuller E. American joint committee on cancer. In Cancer Staging Manual. 6th edn pp. 77–87 New York: Verlag 2002.
BartolazziA Gasbarri A Papotti M Bussolati G Lucante T Khan A Inohara H Marandino F Orlandi F Nardi F Vecchione A Tecce R Larsson O & The Thyroid Cancer Study Group . Application of a immunodiagnostic method for improving preoperative diagnosis of nodular thyroid lesions. Lancet20013571644–1650.
SaggioratoE De Pompa R Volante M Cappia S Arecco F Dei Tos AP Orlandi F & Papotti M. Characterization of thyroid ‘follicular neoplasms’ in fine-needle aspiration cytological specimens using a panel of immunohistochemical markers: a proposal for clinical application. Endocrine-Related Cancer200512305–317.