The authors are grateful to Jane S. Saczynski, PhD (Departments of Internal Medicine and Quantitative Health Sciences, UMASS Medical School) for assistance with statistical analyses and study design.
Although thyroid fine-needle aspiration (TFNA) is an excellent test in evaluating thyroid nodules, there are occasionally false negatives (FN). The clinical impact and pathologic features of FN TFNA is understudied in the peer-reviewed literature.
A cohort of patients with thyroid cancer was separated into those with referring FN TFNA and those with referring true positive (TP) TFNA. Preoperative characteristics, pathologic finding, and clinical outcomes were compared within the 2 groups.
A total of 192 patients with TP TFNA (n = 162) and FN TFNA (n = 30) were included in the study. There were no significant differences in the demographics or length of follow-up of the 2 groups. The FN TFNA group was more likely to have a larger clinical nodule size and experienced a significant delay from initial TFNA to surgery. The FN TFNA group was more likely to be diagnosed with the follicular variant of papillary thyroid cancer (73.3% vs 25.9%, P < .001), less likely to have positive lymph nodes at surgery (6.7% vs 35.8%, P = .001), and more likely to undergo 2-step surgery (30% vs 9.9%, P = .007). Despite the delay in diagnosis, persistent/recurrent or metastatic disease, incidence of aggressive histologic variants, and pT4 disease was not different in the 2 groups.
Thyroid fine-needle aspiration (TFNA) is usually the initial tissue study performed in the evaluation of thyroid nodules, and is used to triage patients to conservative management or surgical management depending on the results of the aspirate. As with any test for malignancy, there is potential for a false negative (FN) TFNA. Patients with thyroid nodules that have concerning clinical or radiographic behavior often undergo excisional biopsies despite negative TFNAs. The risk of malignancy in this highly selected setting varies widely depending on the study but is approximately 8% on average.1-6 This should not be considered the negative predictive value, because most patients with benign thyroid aspirates are followed clinically with good outcomes. A recent study using linear regression modeling estimated the true risk of malignancy to be approximately 2.5% following a benign TFNA.4
The recent Bethesda System for Reporting Thyroid Cytopathology (TBS) is now widely implemented and provides standardized terminology and reporting categories.7 The categories include nondiagnostic (ND), benign, atypical follicular lesion/atypia of uncertain significance (AFL), suspicious/positive for follicular neoplasm (SFN), suspicious for malignancy (SFM), and positive for malignancy (PFM). The risk of malignancy associated with these cytologic categories is fairly well characterized by studies before and after implementation.8-11 The categories can be broadly divided into those that usually trigger surgical excision (SFN, SFM, PFM) and those that usually do not (ND, benign, AFL). In the latter group, clinical and radiographic behavior often prompts the decision to excise the nodule. In this setting, when excision reveals a malignant nodule, the cancer was essentially clinically diagnosed, and thus a case can be made to expand the scope of FN TFNA to include any cancerous thyroid nodule with a referring cytologic diagnosis of ND, benign, or AFL.
False negative results are worrisome to both clinicians and pathologists, because they have the potential to increase patient morbidity due to delay in diagnosis. Clinical and pathologic outcomes in this setting have been the focus of only a few peer-reviewed articles, and thus there is a gap in knowledge of the true impact of FN TFNA. In this study, we examine the pathologic and clinical characteristics and clinical outcomes of patients with histologically confirmed thyroid cancers who had one or more referring FN TFNAs targeting the same nodule.
MATERIALS AND METHODS
The study was approved by the institutional review board of the University of Massachusetts Medical School, Worcester, Massachusetts (Project H-10396). A computer search was used to identify all patients with histologically confirmed thyroid cancer at the University of Massachusetts Medical School, a tertiary care academic medical center, between 2006 and 2010 with one or more precedent TFNAs performed in-house. The cytologic diagnoses were classified according to TBS. Cytologic diagnoses rendered prior to publication of TBS were classified retrospectively based on review of the diagnosis. For the purposes of this study, patients with highest referring cytologic diagnoses of ND, benign, or AFL were classified as FN TFNA. Those with highest referring cytologic diagnoses of FN, SFM, or PFM where classified as true positive (TP) TFNA. Patients with initial ND TFNA and repeat satisfactory TFNA were characterized according to the satisfactory study. Patients with multiple satisfactory TFNA were classified by the highest risk lesion (PFM > SFM > SFN > AFL > benign). Radiographic characteristics of the nodules, histopathologic information including tumor type and stage, and clinical information including follow-up were recorded and tabulated. Follow-up information was retrieved via chart review and included all relevant clinical notes, radiographic studies, and pathologic studies. The interval from TFNA to surgery was measured from the first TFNA targeting the nodule up until surgery. Histologic diagnoses were confirmed to correspond to the lesion targeted by TFNA based on location and size; incidentally identified carcinomas outside of the targeted nodule were not considered in the analysis. The subset of FN TFNAs with referring cytologic diagnoses of benign were reviewed by 2 of the authors (C.L.O. and A.H.F.), and adjudicated diagnoses were rendered. Only the benign TFNA immediately preceding the surgery was reviewed if the patient had multiple FNAs. The reviewers knew the cases were part of this study and were provided patient demographics, but were blinded to all other measured parameters, including subtype of cancer and to the follow-up data. Each review was independent, and if diagnoses were concordant, this became the adjudicated diagnosis. If the reviewers had discordant diagnoses, consensus was reached by reviewing the case together at a multiheaded microscope.
FNA biopsies at The University of Massachusetts were procured either by palpation technique or under ultrasound guidance by endocrinologists or radiologists using a 22- to 25-gauge needle. Between 2 and 6 passes were performed (typically 5). Material from the first 2 to 4 passes was expressed on 1 glass slide, and 1 alcohol-fixed smear per pass was made. Residual material that could not be expressed was rinsed into CytoRich Red fixative (Thermo Scientific, Kalamazoo, Mich). Typically, the last 1 or 2 passes were entirely expressed into CytoRich Red fixative without making smears. The CytoRich Red needle rinse was centrifuged, the supernatant was decanted, and the pellet was resuspended in PreservCyt media (Cytyc, Boxborough, Mass). One ThinPrep slide was prepared and stained with the Papanicolaou technique, using a Sakura Tissue-Tek Prism automated stainer (Sakura Finetek, Torrance, Calif). In some cases, direct smears were stained with Diff-Quik stain (Siemens Diagnostics, Lawrence, Kan) and reviewed for specimen adequacy. If more than 5 mL of solution remained after making the ThinPrep slide or if 4 or more particles were evident in the PreservCyt container, a Cellient cell block (Hologic Corporation, Marlborough, Mass) was prepared.
Demographics and clinical characteristics of the FN TFNA and TP TFNA groups were compared using chi-square tests or 2-tailed Fisher exact test (if n ≤ 5) for ordinal variables and t tests or Wilcoxon-Mann-Whitney test (if nonparametric distribution) for continuous variables. Normality for continuous variables was tested with the Shapiro-Wilk test. Comparisons were considered significantly different at P < .05.
A total of 192 patients with thyroid cancer and referring in-house TFNA targeting the cancerous nodule were identified. Thirty (30 of 192 patients, 15.6%) of these patients had highest referring TFNA diagnoses of either ND, benign, or AFL and formed the FN TFNA group. Of these 30 FN TFNA patients, 11 had referring diagnosis of benign. The remaining 162 (84.4%) patients had highest referring diagnoses of SFN, SFM, or PFM and formed the TP TFNA group. The distribution of referring TFNA diagnoses within the 2 study groups are summarized in Table 1. Table 2 summarizes the 11 cases within the FN TFNA group with highest referring TFNA interpreted as benign. In these 11 cases, adjudicated diagnoses upheld the benign interpretation in 7 cases, 2 cases were adjudicated to AFL, and 2 cases were adjudicated to SFM.
Table 1. Distibution of Referring Thyroid Fine-Needle Aspiration Diagnoses Prior to Resection (n = 192)
AFL indicates atypical follicular lesion/atypia of uncertain significance; SFN, suspicious/positive for follicular neoplasm; ND, nondiagnostic; PFM, positive for malignancy; SFM, suspicious for malignancy.
False negatives (0-2)
True positives (3-5)
Total cases (0-5)
Table 2. Summary of Malignant Cases With Benign Referring Fine-Needle Aspiration (n = 11)
AFL-FN indicates atypical follicular lesion–exclude follicular neoplasm; A-NED, alive with no evidence of disease during follow-up; A-WED, alive with evidence of nonlocal recurrence or distant metastasis during follow-up; Dem, demographics; FC, follicular carcinoma; FCMI, minimally invasive follicular carcinoma; FU, follow-up; FVPTC, follicular variant of papillary thyroid carcinoma; N/A, not applicable; NFMC, negative/benign; S-PTC, suspicious for papillary thyroid carcinoma; Size-P, pathologic size of cancer; Size-R, radiographic size of nodule.
Patient characteristics, histologic outcomes, and clinical outcomes are summarized in Table 3. There were no significant differences between the demographics and length of follow-up between the FN TFNA (female:male ratio, 2.33; mean age, 47.9 years; mean follow-up, 28.2 months) and TP TFNA (female:male ratio, 2.76; mean age, 48.2 years; mean follow-up, 28.8 months) groups. The clinical size of the tumor was significantly larger in the FN TFNA group (mean, 4.3 cm vs 2.4 cm, P = .047; median, 3.3 cm vs 1.8 cm, P = .002). The time from initial TFNA to surgery was significantly longer in the FN TFNA group (mean, 14.3 months vs 1.9 months, P < .001; median, 10 months vs 2 months, P < .001).
Table 3. Comparison of False Negative Group to True Positive Group (n = 192)
FN (n = 30)
TP (n = 162)
Indicates nonparametric distribution of data, except the Wilcoxon-Mann-Whitney test P value of the median.
Data available for 113 patients (15 in FN group and 97 in TP group), clinical size is the size according to radiographic study.
Both categories include Hürthle-type of follicular cancers.
Alive with histologic or radiographic evidence of recurrent disease in nonlocal lymph nodes or distant metastatic disease during period of follow-up.
FC indicates follicular carcinoma; FCMI, minimally invasive follicular carcinoma; FN, false negative; FNA, fine-needle aspiration; FVPTC, follicular variant of papillary thyroid carcinoma; MC, medullary carcinoma; SD, standard deviation; PDCA, poorly differentiated cancer including insular and anaplastic carcinoma; PTC, papillary thyroid carcinoma; TP, true positive.
The FN TFNA group was significantly more likely to have a final diagnosis of follicular variant of papillary thyroid carcinoma (73.3% of cases vs 25.9%, P ≤ .001) and less likely to be diagnosed with conventional or mixed type of papillary thyroid carcinoma (6.7% vs 56.8%, P < .001). The FN TFNA group was significantly less likely to be American Joint Commission on Cancer stage pT1 (26.7% vs 55.6%, P = .004) and significantly more likely to be stage T2 (46.7% vs 23.5%, P = .009). Lymph node involvement at the time of surgery was significantly less in the FN TFNA group (6.7% vs 35.8%, P = .001). The pathologic size of the cancers was significantly larger in the FN TFNA group (mean, 3.5 cm vs 2.3 cm, P < .001; median, 3.2 cm vs 1.7 cm, P < .001). Patients in the FN TFNA group were significantly more likely to undergo 2-step surgery for completion thyroidectomy (30% vs 9.9%, P = .007).
Among patients with follow-up information, the FN TFNA group had 24 of 25 (96%) of patients alive with no evidence of recurrence during the follow-up period and 1 of 25 (4%) patients was alive with evidence of recurrent (excluding regional lymph nodes) and/or metastatic disease during the follow-up period. There were no patient deaths in the FN TFNA group. Within the TP TFNA group, 130 of 146 (89%) of patients were alive with no evidence of disease during the follow-up period and 13 of 146 (9%) were alive with evidence of recurrent (excluding regional lymph nodes) and/or metastatic disease during the follow-up period. Three deaths occurred in the TP FNA group, 2 of which were related to thyroid cancer and the third of which was not related to thyroid cancer. The TP TFNA results were obtained on the first TFNA targeting the malignant nodule in all of the 13 patients alive with evidence of disease and in both patients who died of thyroid cancer. There were no significant differences in outcomes between the FN TFNA and TP TFNA groups. Figure 1 illustrates one of the FN TFNAs and its follow-up.
False negative results in any test for cancer have the potential to increase patient morbidity due to the delay in diagnosis. TFNA has emerged as the initial tissue study in the evaluation of thyroid nodules and is an excellent test for triaging patients to surgical intervention or conservative follow-up. TBS for reporting thyroid cytology recognizes equivocal categories to increase the sensitivity for thyroid cancer. Despite these equivocal categories, on occasion, nodules with referring TFNA diagnoses of benign, ND, or mildly atypical are ultimately discovered to be thyroid cancer. It is difficult to know what the true FN rate is for TFNA, because most nodules with benign cytology are not surgically excised. The risk of malignancy in the selected setting of patients who undergo excision of a thyroid nodule despite one or more benign FNAs is variable depending on the study, but averages approximately 8%.1-6 Most would agree the true risk of malignancy is much lower in patients with benign TFNA, being closer to 1% to 3%.4
Although there are good data regarding the risk of malignancy in nodules with benign cytology and atypical clinical behavior and a reasonably good estimate of the true negative predictive value of TFNA, the clinical impact of FN TFNA is understudied. To our knowledge, only one other series has examined the impact of FN TFNA by comparing outcomes to patients within the same population diagnosed with thyroid cancer via TFNA (ie, TP TFNAs).12 Yeh and colleagues reported a series of thyroid cancers diagnosed with referring benign or inadequate TFNA and compared them to thyroid cancers diagnosed with referring positive or suspicious TFNA. They reported that the FN TFNA group experienced, on average, a 2-year delay in surgery and were twice as likely to have persistent disease on follow-up. In the FN TFNA group, 4 of 12 were pT4 at the time of diagnosis, 1 of 12 was an aggressive histologic cancer subtype (insular cancer), and 3 of 12 had disseminated or aggressive local disease. The group of Yeh and colleagues concludes that FN TFNA adversely affects outcomes in patient with thyroid cancer.
The results of our study add to the literature in several ways. Contrasting our findings to those of Yeh et al, in our cohort of FN TFNA, every case was histologically low grade and most of the cases were either the follicular variant of papillary thyroid carcinoma or minimally invasive follicular carcinoma. Aggressive local disease (ie, pT4 disease) was not present in any of our cases, and 24 of 25 patients with FN TFNA and follow-up data are alive without evidence of disease at latest follow-up. Similar to the findings of Yeh et al, time to surgery and thus diagnosis was significantly delayed in our FN TFNA cohort, and the FN TFNA group was more likely to be subjected to a 2-step surgery. The 1 patient with aggressive disease in our FN TFNA group was treated with total thyroidectomy without tremendous delay due to atypical clinical behavior (Table 2, case #3). Thus, this particular FN TFNA did not adversely affect patient care, and we consider this case a “near miss.” In summary, no patient in our series had adverse outcome related to the FN TFNA.
Differences in methodology may explain the differences in our results compared with those of Yeh and colleagues. In their series, it is noted that some of the patients were referred for surgery with referring TFNA(s) occurring at outside hospitals. In the current study, only thyroid cancer patients with in-house referring TFNA(s) were included. Thus, there may be a referral bias in the cohort of Yeh et al, based on selective referral of patients in need of more complicated surgeries and, by extension, more aggressive disease. Alternatively, the differences in pathology may be the result of the difference in delay (14 months versus 2 years) of surgery. If so, careful clinical evaluation at 1 year and not at a later time may be a critical part of ensuring the success of TFNA.
Our adjudicated diagnoses revealed 4 of 11 cases to be interpretative errors. Of the interpretative errors, 2 cases were major interpretative errors with adjudicated diagnoses of “suspicious for malignancy,” which usually triggers surgery. The other 2 interpretative errors are minor and were adjudicated to “atypical follicular lesion,” which may or may not trigger surgery depending on the clinical setting. The fact that most of the benign cases within our FN TFNA were noninterpretative errors supports the idea that many malignant thyroid nodules are sufficiently heterogeneous that an adequate TFNA may only contain cytologically and architecturally normal cells. Other investigators have shown that the pathologic size of the malignancy in thyroid cancer was significantly smaller than the clinical size of the nodule in FN TFNAs, which also supports this notion.13
The fact that the follicular variant of papillary thyroid carcinoma was strongly associated with the FN TFNA group is not surprising. This variant can be difficult to recognize due to subtle cytologic features, and other investigators have noted the challenge of recognizing this type of cancer on TFNA.14, 15 It is also worth noting the significant intraobserver variability and the recognized problem of defining minimal diagnostic criteria for this diagnosis.16 The significantly larger clinical size of the nodule and the larger pathologic size seen in the FN TFNA group are probably related to the fact that large clinical size is one of the criteria that will lead to surgical intervention. This same clinical consideration offers explanation for the differences seen in pT1 and pT2 status between the FN TFNA and TP TFNA groups. Although there were no statistically significant differences between pT4 statuses, incidence of poorly differentiated carcinoma, and incidence of medullary carcinoma in our 2 study groups, it is worth pointing out that no such cases occurred in our FN TFNA group.
Our study has a few limitations. The study population was of relatively small size. Also, our institution performs a high volume of TFNA and thyroid surgical pathology studies annually, and challenging cases are reviewed regularly at multidisciplinary conferences with referring clinicians and surgeons, many of whom have a large focus of their practice on nodular thyroid disease. Thus, our findings may only be generalizable to similar settings. Additional studies examining FN TFNA in different settings would be useful to further define the clinical impact of FN TFNA.
In conclusion, although TFNA fails to identify a significant subset of patients with thyroid cancer, the clinical consequence of FN TFNA in our high-volume center is minimal. Thyroid cancers that elude detection with TFNA, in our experience, are low grade and most are cured despite the false negative cytologic diagnosis. Our results demonstrate that outcomes are not significantly different in FN TFNAs compared with TP TFNAs, despite delayed treatment. Our results reemphasize the important concept that TFNA is just one consideration in the decision algorithm of managing nodular thyroid disease and that best patient care incorporates clinical, radiographic, and cytologic findings.