Predictors of Bilateral Disease in Pediatric Differentiated Thyroid Cancer

Abstract

Context

Total thyroidectomy is recommended for children with papillary thyroid carcinoma, partly because of a high prevalence of bilateral disease. Identifying characteristics that predict bilateral disease might identify candidates for more limited surgery.

Objective

Investigate associations of preoperative or histopathological characteristics with bilateral disease in children with differentiated thyroid cancer.

Methods

Retrospective cohort study (1998-2020) at 2 academic hospitals. Patients <19 years who underwent total thyroidectomy for differentiated thyroid cancer were included. Clinical, sonographic, and histopathological characteristics were evaluated. The presence of bilateral disease on histopathology was assessed by univariable analysis and multivariable logistic regression.

Results

One hundred and fifteen subjects were analyzed (90% with papillary carcinoma). Median (range) age at diagnosis was 15.0 (8.1-18.9) years. Bilateral disease was present in 47/115 subjects (41%). Bilateral disease was associated with solid parenchyma, calcifications, irregular margins, and abnormal lymph nodes detected by ultrasound, Bethesda class V/VI cytology, papillary histology, tumor multifocality in the primary lobe, extrathyroidal extension, lymphovascular invasion, and nodal metastases. In multivariable analysis, only multifocality in the primary lobe was independently associated with bilateral disease (OR 7.61, 95% CI 2.44-23.8, P < .001). Among clinically node-negative subjects with papillary carcinoma who did not have tumor multifocality in the primary lobe, bilateral disease was present in 5/32 (16%).

Conclusions

In children with differentiated thyroid cancer, tumor multifocality in the primary lobe is associated with bilateral disease and should prompt consideration of completion thyroidectomy after initial lobectomy. Clinically node-negative children with tumors that are unifocal in the primary lobe have a low likelihood of contralateral disease.

Thyroid cancer is the most common endocrine cancer in children, with papillary thyroid carcinoma (PTC) accounting for the majority of cases (1, 2). Current guidelines recommend total thyroidectomy as the initial surgical approach for children with papillary thyroid cancer (3) based on the significant rate of bilateral disease in children (30-40%) (3, 4) and the lower rate of disease recurrence after total thyroidectomy than less complete resection (5). In adults with low-risk thyroid cancer, recent evidence and guidelines support that more limited surgery is adequate (6), but whether the same is true in children remains unclear. Given the risks of thyroid surgery in children (7), identifying pediatric patients who might safely be managed with less extensive surgery is a high priority. We sought to identify characteristics of pediatric patients, thyroid nodules, and thyroid cancers that might predict unilateral thyroid cancer that may be amenable to limited surgical resection (lobectomy). Specifically, we sought to address 2 clinical questions regarding pediatric thyroid cancer: First, whether any preoperative demographic or sonographic characteristics can predict the presence of bilateral disease to guide the initial operative approach, and, second, whether any preoperative or histopathological characteristics of the primary tumor can predict the presence of contralateral disease that might require completion thyroidectomy after initial lobectomy.

Materials and Methods

Subjects

We reviewed the medical records of all patients with non-medullary differentiated thyroid cancer diagnosed before 19 years of age at Boston Children’s Hospital or Brigham and Women’s Hospital between January 1998 and March 2020. Patients were included if they had a preoperative ultrasound performed in our multidisciplinary pediatric thyroid clinic, underwent complete resection of the thyroid gland (total thyroidectomy or lobectomy followed by completion thyroidectomy), and if surgical histopathology of the entire thyroid was reviewed at 1 of our 2 institutions. Preoperative ultrasounds were performed by radiologists with expertise in thyroid ultrasonography. Ultrasound-guided fine-needle aspiration (FNA) cytology was interpreted by a Brigham and Women’s Hospital cytopathologist and reported according to the Bethesda System for Reporting Thyroid Cytopathology (8). Although the beginning of the study period predates the formal introduction of the Bethesda system, all cytopathologists employed diagnostic criteria and terminology used by the Bethesda system. For patients with thyroid cancer, therapeutic lymph node dissection was performed for known or clinically suspected lymph node metastasis; prophylactic lymph node dissection was not performed.

Data Collection

Demographic and clinical characteristics of subjects were abstracted from the medical record. Preoperative sonographic features of each thyroid cancer included size, cystic content, echogenicity, borders, calcifications, the presence of other nodules ≥5 mm in diameter, and abnormal-appearing lymph nodes. Lymph nodes were defined as abnormal if they had calcifications, abnormal shape, no visible hilum, or abnormal echotexture. FNA cytology of each nodule was recorded. Histopathological features included tumor size, histological type (papillary or follicular), laterality (unilateral vs bilateral), focality in the primary tumor lobe (unifocal or multifocal), and the presence of lymphovascular invasion, extrathyroidal extension, and pathological lymph node metastases. For this study, cancer multifocality was defined as either (1) more than 1 focus of thyroid cancer in the lobe containing the primary thyroid carcinoma, or (2) thyroid cancer involving the entire primary lobe (as is commonly observed in the diffuse sclerosing variant of PTC); the presence or absence of tumor in the contralateral lobe was not considered in the definition of multifocality. This definition was chosen a priori to assess the association between focality in the primary lobe (which can be evaluated in a lobectomy specimen) and the presence of contralateral disease that might be an indication for completion thyroidectomy. Because lymphatic and vascular invasion were not distinguished in some older pathology reports, the composite variable lymphovascular invasion was used. Ultrasound images were reviewed by a radiologist (D.M.R.) and original histopathology was reviewed by a pathologist (J.A.B. or M.L.H.) when needed to clarify clinical reports. When specific data elements for a given patient were unavailable in the medical record and primary materials were not available for review, these data elements were excluded from analysis. Study data were managed using REDCap Software Version 7.6.1 (9).

Statistical Analysis

The primary outcome was the presence of bilateral disease, defined as at least 1 focus of thyroid cancer in each lobe. Associations of subject and cancer characteristics with the primary outcome were evaluated by chi-squared analysis for categorical variables or Mann–Whitney U-test for continuous variables, and odds ratios were calculated using univariable logistic regression. Variables that were associated with the outcome in univariable analysis (P < .2) were included in multivariable logistic regression analysis. Because follicular carcinoma (FTC) is rarely associated with bilateral disease, we also performed a sensitivity analysis excluding cases of FTC. However, all differentiated thyroid cancers (PTC and FTC) were included in the primary analysis assessing preoperative factors associated with bilateral disease because the histological tumor type may not be known preoperatively.

Because the decision to perform initial total thyroidectomy in patients with clinical lymph node involvement on preoperative evaluation (cN1) may be driven by the anticipated need for postoperative radioiodine therapy rather than concern for bilateral disease, we performed a secondary analysis excluding subjects with preoperative cN1 disease. Because of the smaller size of this secondary analysis, only variables that were associated with the outcome at a significance level of P < .05 in univariable analysis were included in the multivariable analysis. For all analyses, a 2-tailed P < .05 was considered statistically significant. All statistical analyses were performed using SAS version 9.4 (SAS Institute). This study was approved by the Institutional Review Boards of Boston Children’s Hospital and Brigham and Women’s Hospital.

Results

During the study period, 115 pediatric patients with differentiated thyroid cancer met inclusion criteria. Most patients were female (88/115, 77%), and the median (range) age at diagnosis was 15.0 (8.1-18.9) years. A history of prior malignancy was present in 8 subjects (7%), and 7 subjects (6%) had prior radiation exposure (total body irradiation, range 12-14 Gy, n = 5; external beam radiation, range 15.0-23.4 Gy, n = 2). Fifteen subjects (13%) had a significant family history of thyroid cancer and 3 subjects (3%) had a known monogenic risk factor for thyroid cancer, including a pathogenic variant in PTEN (n = 2) or APC (n = 1). Preoperative evaluation revealed clinical lymph node involvement (cN1) in 59/115 cases (51%). Ninety subjects (78%) underwent total thyroidectomy, and 25/115 (22%) underwent lobectomy followed by completion thyroidectomy. PTC was present in 104 cases (90%) and FTC was present in 11 cases (10%). PTC histological subtypes included classical (54/104, 52%), diffuse sclerosing variant (23/104, 22%), follicular variant (20/104, 19%), solid variant (3/104, 3%), and other variants (4/104, 4%).

In the study cohort, 47/115 subjects (41%) had bilateral disease on histopathology. The associations of bilateral disease with subject, preoperative, and postoperative characteristics are shown in Table 1. The presence of bilateral disease was not associated with any subject characteristic, including sex, age, or medical history. Preoperative factors statistically associated with bilateral disease included solid parenchyma, irregular margins, calcifications, and abnormal-appearing lymph nodes on ultrasound, and FNA cytology of Bethesda category V or VI. The presence of additional thyroid nodules ≥5 mm on preoperative ultrasound (either ipsilateral or contralateral to the primary tumor) was not associated with bilateral thyroid cancer. In addition, the presence of additional ipsilateral thyroid nodules on ultrasound was not associated with tumor multifocality in the primary lobe on histopathology (30% vs 25%, P = .34). All of the postoperative characteristics analyzed were associated with bilateral disease, including papillary histology, tumor multifocality in the primary lobe, extrathyroidal extension, lymphovascular invasion, and pathological lymph node metastases (pN1).

Table 1.

Characteristics of the study cohort and associations between subject and thyroid characteristics and bilateral thyroid cancer

	Total	Unilateral disease	Bilateral disease	Univariable		Multivariable
				OR (95% CI)	P	OR (95% CI)	P
Subject characteristics	n = 115	n = 68	n = 47
Female sex	88 (77)	50 (74)	38 (81)	1.52 (0.62-3.76)	.36
Age at diagnosis, y	15.0 (8.1-18.9)	15.1 (8.1-18.9)	14.8 (9.3-18.6)	0.97 (0.84-1.12)	.68
Thyroid disease	30 (26)	15 (22)	15 (32)	1.65 (0.72-3.83)	.24
Prior malignancy	8 (7)	4 (6)	4 (9)	1.49 (0.35-6.27)	.59
Radiation exposure	7 (6)	2 (3)	5 (11)	3.92 (0.73-21.2)	.11	6.93 (0.94-51.2)	.06
Genetic risk	3 (3)	2 (3)	1 (2)	0.71 (0.06-8.14)	.78
Family history of thyroid neoplasia	15 (13)	10 (15)	5 (11)	0.69 (0.22-2.17)	.53
Preoperative nodule characteristics
Diameter, mm	27 (3-78)	25 (3-64)	27 (8-78)		.30
Diameter ≥20 mm	82 (71)	48 (71)	34 (72)	1.09 (0.48-2.49)	.84
Solid parenchyma	89 (77)	45 (66)	44 (94)	7.50 (2.1-26.8)	.002	2.63 (0.34-20.2)	.35
Calcifications (n = 111)	82 (74)	40/64 (63)	42/47 (89)	5.04 (1.75-14.5)	.003	0.91 (0.09-9.21)	.85
Irregular margins (n = 110)	65 (59)	29/63 (46)	36/47 (77)	3.84 (1.66-8.86)	.002	2.00 (0.43-9.28)	.93
Taller than wide (n = 90)a	21 (22)	11/54 (20)	10/36 (28)	1.5 (0.56-4.03)	.42
Hypoechoic (n = 112)	75 (67)	40/64 (62)	35/47 (74)	1.6 (0.7-3.66)	.27
Ipsilateral nodule(s) present (n = 101)	27 (27)	14/57 (25)	13/44 (30)	1.49 (0.65-3.43)	.34
Contralateral nodule(s) present (n = 101)	14 (14)	7/57 (12)	7/44 (16)	1.29 (0.43-3.84)	.65
Clinical lymph node disease (cN1)	59 (51)	24 (35)	35 (74)	5.34 (2.34-12.2)	<.001	0.65 (0.13-3.28)	.61
Bethesda V-VI cytologyb	84 (73)	41 (60)	43 (91)	7.09 (2.28-22.1)	<.001	1.46 (0.23-9.32)	.69
Postoperative cancer characteristics
Papillary carcinoma	104 (90)	57 (84)	47 (100)	—	.007	c	c
Multifocality in primary lobe (n = 109)	46 (42)	16/68 (24)	30/41 (73)	8.86 (3.64-21.6)	<.001	7.61 (2.44-23.8)	<.001
Extrathyroidal extension	38 (33)	15 (22)	23 (49)	3.94 (1.51-7.61)	.003	1.25 (0.38-4.08)	.71
Lymphovascular invasion	75 (65)	38 (56)	37 (79)	2.92 (1.25-6.81)	.01	0.64 (0.14-3.05)	.58
Pathological lymph node disease (pN1)	69 (60)	28 (41)	41 (87)	9.77 (3.65-26.1)	<.001	5.08 (0.58-44.5)	.14

Data reported as number (%) or median (range). Denominators are provided for characteristics for which data were not available for all subjects. Characteristics associated with bilateral disease in univariable analysis (P < .2) were included in the multivariable analysis.

^aSubjects not included in analysis if no discrete nodule was present or the entire lobe was involved.

^bCompared with Bethesda I-IV cytology.

^cVariable not included in multivariable analysis because no subject with follicular carcinoma had bilateral disease.

In multivariable analysis, only tumor multifocality in the primary lobe remained statistically associated with bilateral disease (OR 7.61, 95% CI 2.44-23.8, P < .001). Bilateral disease was present in 30/46 (65%) subjects with multifocality in the primary lobe but in 16/63 (25%) subjects without multifocality. The odds of bilateral disease were increased in subjects with a history of radiation exposure (OR 6.93, 95% CI 0.94-51.2, P = .06) and pathological lymph node disease (pN1, OR 5.08, 95% CI 0.58-44.5, P = .14), but these associations were not statistically significant. Among subjects with a history of radiation exposure, the median (range) radiation dose was similar in patients with or without bilateral disease (14 [12-23.4] Gy vs 13.8 [13.5-14] Gy, P = .59).

Bilateral disease was present only in subjects with PTC; because no subject with FTC had bilateral disease, a specific odds ratio could not be calculated and this variable was not included in the multivariable analysis. When all analyses were repeated after excluding subjects with FTC (n = 11), similar results were obtained. Among subjects with PTC who did not have multifocal tumor in the primary lobe, bilateral disease was present in 11/52 (21%).

A secondary analysis was performed that included only subjects without clinical lymph node involvement on preoperative evaluation (cN0, n = 56, Table 2). In univariable analyses, bilateral disease was associated with radiation exposure, smaller tumor size, Bethesda V-VI cytology, multifocality in the primary lobe, and pathological lymph node disease. In multivariable analysis, only multifocality in the primary lobe remained significantly associated with bilateral disease (OR 14.1, 95% CI 1.61-123.3, P = .02). Among cN0 subjects who did not have tumor multifocality in the primary lobe, bilateral disease was present in 5/43 (12%) subjects overall and in 5/32 (16%) subjects with PTC. In cN0 subjects, the odds of bilateral disease were increased with pathological lymph node disease (OR 7.30, 95% CI 0.91-58.4, P = .06) and were inversely related to tumor size (OR 0.16, 95% CI 0.02-1.1, P = .06), but these associations were not statistically significant. All patients in this analysis with radiation exposure (n = 3) had bilateral disease, so a specific odds ratio could not be calculated and this variable was not included in the multivariable analysis.

Table 2.

Characteristics of subjects with no preoperative evidence of clinical lymph node involvement (cN0), and associations between subject and thyroid characteristics and bilateral thyroid cancer

	Total	Unilateral Disease	Bilateral Disease	Univariable		Multivariable
				OR (95% CI)	P	OR (95% CI)	P
Subject characteristics	n = 56	n = 44	n = 12
Female sex	46 (82)	35 (80)	11 (92)	2.82 (0.32-24.9)	.35
Age at diagnosis, y	15.4 (8.1-18.9)	15.4 (8.1-18.9)	15.3 (9.3-18.6)	0.94 (0.73-1.21)	.63
Thyroid disease	14 (25)	10 (23)	4 (33)	1.70 (0.42-6.84)	.46
Prior malignancy	3 (5)	1 (2)	2 (17)	8.60 (0.71-105)	.09
Radiation exposure	3 (5)	0 (0)	3 (25)	-	.008	c	c
Genetic risk	3 (5)	2 (5)	1 (8)	1.91 (0.16-23.0)	.62
Family history of thyroid neoplasia	11 (20)	8 (18)	3 (25)	1.50 (0.33-6.82)	.60
Preoperative nodule characteristics
Diameter, mm	24 (8-32)	26 (8-32)	14 (8-32)		.003
Diameter ≥20 mm	35 (62)	32 (73)	3 (25)	0.13 (0.03-0.54)	.005	0.16 (0.02-1.1)	.06
Solid parenchyma	33 (59)	23 (52)	10 (83)	4.57 (0.90-23.3)	.07
Calcifications (n = 52)	21 (43)	15/40 (38)	6/12 (50)	1.89 (0.51-7.00)	.34
Irregular margins (n = 52)	21 (43)	14/40 (35)	7/12 (58)	2.50 (0.67-9.37)	.17
Taller than wide (n = 45)a	9 (20)	6/35 (17)	3/10 (30)	1.71 (0.35-8.37)	.51
Hypoechoic (n = 51)	35 (69)	25/39 (64)	10/12 (83)	2.59 (0.5-13.5)	.26
Ipsilateral nodule(s) present (n = 49)	11 (22)	9/39 (23)	2/12 (17)	0.76 (0.14-4.08)	.74
Contralateral nodule(s) present (n = 49)	9 (16)	6/39 (15)	3/12 (25)	2.06 (0.43-9.84)	.37
Bethesda V-VI cytologyb	30 (54)	19 (43)	11 (92)	14.47 (1.72-122)	.01	6.23 (0.45-85.5)	.17
Postoperative cancer characteristics
Papillary carcinoma	45 (80)	33 (75)	12 (100)	—	.10
Multifocality in primary lobe (n = 54)	11 (20)	6/44 (14)	5/10 (50)	6.33 (1.40-28.6)	.02	14.1 (1.61–123.3)	.02
Extrathyroidal extension	6 (11)	3 (7)	3 (25)	4.56 (0.79-26.4)	.09
Lymphovascular invasion	26 (46)	19 (43)	7 (58)	1.84 (0.51-6.71)	.35
Pathological lymph node disease (pN1)	13 (23)	6 (14)	7 (58)	8.87 (2.12-37.2)	.003	7.30 (0.91–58.4)	.06

Data reported as number (%) or median (range). Denominators are provided for characteristics for which data were not available for all subjects. Characteristics associated with bilateral disease in univariable analysis (P < .05) were included in the multivariable analysis

^aSubjects not included in analysis if no discrete nodule was present or the entire lobe was involved. ^bCompared to Bethesda I-IV cytology.

^cVariable not included in multivariable analysis because no subject with radiation exposure had unilateral disease.

In the full study cohort, tumor size was not associated with bilateral disease (P = .84). However, in the secondary analysis of cN0 subjects, tumors ≥20 mm were less likely to be bilateral than were smaller tumors (3/35 [9%] vs 9/21 [43%], P = .005). To further investigate this relationship we performed a post hoc analysis of cN1 subjects. This demonstrated the opposite finding that among cN1 subjects, tumors ≥20 mm had a higher rate of bilateral disease than did smaller tumors (31/47 [66%] vs 4/12 [33%], P = .05).

Discussion

In this cohort of children with differentiated thyroid cancer, the prevalence of bilateral disease was high (41%) and consistent with that of other studies (30-40%) (3, 4). This high prevalence of bilateral disease has served as a rationale for current consensus guidelines recommending total thyroidectomy for all children with PTC (3). This approach is also supported by retrospective data demonstrating improved long-term outcomes in children treated with total thyroidectomy compared with less complete surgery (5). However, in adults, lobectomy is now recognized to be adequate treatment for many low-risk thyroid cancers (6, 10). Emerging data suggest that a similar approach may be viable in children (11), but additional outcome data are needed. In the interim, concern about the frequent presence of bilateral disease remains a significant factor driving clinical decisions to pursue total thyroidectomy in children with thyroid cancer. Therefore, we sought to identify preoperative or histopathological characteristics that may guide surgical decision-making by identifying patients at particularly high or low risk of bilateral disease.

The first clinical question we sought to address is whether preoperative factors associated with bilateral disease may help guide decisions about the extent of initial surgery. It is important to note that this clinical scenario refers to patients in whom the diagnosis of thyroid cancer is known or highly likely preoperatively, such that total thyroidectomy is an appropriate option. For patients at lower risk of malignancy (eg, patients with indeterminate cytology), diagnostic lobectomy remains appropriate. In our cohort, no preoperative subject or nodule characteristic had a statistically significant, independent association with bilateral disease. However, radiation exposure was associated with a nearly 7-fold increased risk of bilateral disease that was of borderline statistical significance, suggesting that a significant association might be observed in a larger cohort. In addition, its high positive predictive value for bilateral disease (5/7, 71%) suggests that a history of radiation exposure could reasonably prompt consideration of total thyroidectomy if thyroid cancer is known or highly likely to be present based on preoperative ultrasound and cytology.

The second relevant clinical question was whether any histopathological tumor features associated with bilateral disease might inform the decision to resect the contralateral lobe in a patient diagnosed with thyroid cancer after initial lobectomy (eg, for a cytologically indeterminate nodule). As might be expected, multiple aggressive histological features were associated with bilateral disease in univariable analysis. However, multivariable analysis revealed that only tumor multifocality in the primary lobe was independently associated with bilateral disease.

The clinical significance of thyroid cancer multifocality remains uncertain. Although multifocality is associated with more extensive disease, studies in adults with thyroid cancer have reached conflicting conclusions about whether multifocality is an independent predictor of disease recurrence, progression, or mortality (12-16). In many of these studies, the definition of multifocality is inconsistent, potentially comprising multiple tumor foci in the same lobe as the primary tumor or additional tumor foci anywhere in the gland, including in the contralateral lobe. We chose to define multifocality in this study as multiple foci in the primary tumor lobe or tumor occupying the entire primary lobe, allowing this characteristic to be assessed in a lobectomy specimen and thereby utilized for clinical decision-making regarding completion thyroidectomy. In our cohort, multifocality in the primary lobe was associated with nearly 8-fold increased odds of bilateral disease and had a positive predictive value of 65% for the presence of disease in the contralateral lobe. Perhaps more clinically relevant, however, is the finding that absence of tumor multifocality in the primary lobe was associated with unilateral disease in 79% of pediatric PTCs overall and in 84% of PTCs without clinical lymph node involvement (cN0). Moreover, among children presenting with clinically low-risk cN0 PTCs, bilateral disease was present in 45% (5/11) of subjects with multifocality in the primary lobe but only 16% (5/32) of subjects without multifocality. The latter rate of bilateral disease is only slightly higher than that reported in adult patients with PTC (5-13%), in whom lobectomy is accepted to be adequate treatment in the absence of aggressive tumor features (17, 18). These findings suggest that concern for bilateral disease alone may not justify total thyroidectomy as the universal approach to thyroid cancer surgery in children and that additional factors, including tumor multifocality, should be considered in decisions about the extent of surgery.

No FTC in our cohort was bilateral. Although this finding might have been expected, we chose to include FTCs in our primary analysis because the histological tumor type may not be known preoperatively, and part of our goal was to identify preoperative features that might guide decisions about the extent of initial surgery. Nevertheless, a sensitivity analysis excluding FTCs showed similar results. For FTCs, the decision of whether to complete thyroidectomy after initial lobectomy will likely be driven by the extent of vascular invasion, which is associated with an increased risk of adverse outcomes and the associated need for adjunctive radioiodine therapy, rather than by concern for potential contralateral disease.

Interestingly, preoperative clinical lymph node involvement (cN1) was not independently associated with bilateral disease in our cohort. This is consistent with the finding of a large Japanese study that lobectomy may be associated with a low risk of disease recurrence or mortality in some children with thyroid cancer even in the presence of nodal metastases (11). Nevertheless, preoperative cN1 status may be an independent indication for initial total thyroidectomy if adjunctive radioiodine therapy is anticipated postoperatively. Therefore, we conducted a secondary analysis to determine whether our findings remained consistent in the subgroup of patients without preoperative lymph node disease (cN0). These results were similar, with radiation exposure and multifocality in the primary lobe being the only factors independently associated with bilateral disease. Thus, in cN0 patients who undergo lobectomy—including for cytologically indeterminate nodules—tumor multifocality in the primary lobe should raise concern for potential residual disease in the contralateral lobe.

The presence of contralateral nodules on preoperative ultrasound was not associated with bilateral disease. Although perhaps surprising, this finding is consistent with data from adults demonstrating that the presence of contralateral nodules on ultrasound does not significantly increase the risk of contralateral malignancy after lobectomy for thyroid cancer (18). Similarly, in our cohort the presence of ipsilateral nodules on ultrasound was not associated with either bilateral disease or tumor multifocality in the primary lobe. These findings indicate that sonographic nodularity is distinct from, and not necessarily predictive of, histopathological malignancy in children with thyroid cancer. Therefore, the presence of sonographically nonsuspicious ipsilateral or contralateral nodules should not prompt total thyroidectomy solely based on concern for bilateral disease. Nevertheless, contralateral findings that are specifically associated with malignancy—such as sonographically suspicious nodules or microcalcifications suggestive of the diffuse sclerosing variant of PTC—should be evaluated and managed appropriately.

In this analysis, pathological lymph node disease (pN1) was not statistically associated with bilateral disease. This finding contrasts with that of another recent study that did observe an association between these features (4). However, the odds ratio for this association in our cohort was high and the 95% CI was wide (OR 5.08, 95% CI 0.58-44.5), suggesting that a statistically significant association might be apparent in a larger cohort.

The associations between tumor size and bilateral disease in our cohort were somewhat unexpected. In a previous study, tumor size ≥20 mm was found to be associated with bilateral disease (4). This association was not observed in our full cohort but was observed among the subgroup of subjects with preoperative cN1 disease. Interestingly, the opposite was true of cN0 subjects, in whom smaller tumors <20 mm were more likely to be bilateral. A possible reason for these findings might be inherently discrepant behavior between tumors that are prone to local metastasis and those that are not. For example, aggressive tumors that present as cN1 disease may tend to spread both intrathyroidally and to local lymph nodes in proportion to their size, whereas more indolent tumors or FTCs that grow to large sizes without metastasizing to local lymph nodes may be more prone to remain localized within the thyroid as well. The clinical implications of this finding are unclear, as it seems counterintuitive to suggest that cN0 patients with smaller thyroid cancers should be considered for total thyroidectomy, whereas those with larger cancers may require only lobectomy. However, given the limited sample size and the lack of independent association in multivariable analysis, additional studies are needed to elucidate the clinical relevance of these findings relating nodule size to risk of bilateral disease.

A major strength of this study is its use of multivariable analysis to determine features that are independently associated with the presence of bilateral disease in a relatively large number of pediatric thyroid cancers. Another recent study in children with PTC found that bilateral disease was associated not only with multifocality/multinodularity on ultrasound but also with tumor size ≥2 cm, extrathyroidal extension, lymphovascular invasion, positive central neck lymph nodes, extranodal extension, and the diffuse sclerosing subtype of PTC (4). However, multivariable analysis was not performed in this prior study, which is significant because in our multivariable analysis many of these features were not independently associated with bilateral disease. Indeed, other reports have found that aggressive tumor features are common in patients with multifocal disease (12), which was the only feature to be independently associated with bilateral disease in our cohort.

Our study has several limitations, primarily related to its retrospective design. Despite its relatively large number of pediatric thyroid cancers, this cohort may not be of sufficient size to detect small effects. Additionally, based on the data available over the study period, we were unable to analyze some other features of potential interest, such as degree of extrathyroidal extension, number or size of lymph node metastases, or presence of extranodal extension.

A final limitation is that the primary outcome of this study (bilateral thyroid cancer) is not a clinical outcome such as disease recurrence or mortality. It remains uncertain whether bilateral disease is associated with adverse outcomes in children with thyroid cancer or whether the removal of potential contralateral disease improves outcomes and therefore warrants total thyroidectomy over lobectomy. Studies in adults with thyroid cancer indicate that rates of disease recurrence and mortality are low in patients undergoing lobectomy for low-risk thyroid cancer, either with or without tumor multifocality (18, 19). However, whether the same is true in children is unknown, particularly in light of the differences between children and adults in thyroid cancer genetics and the post-treatment lifespan during which they remain at risk. One study has suggested that lobectomy may be sufficient for the treatment of children with low-risk thyroid cancer (11), but these findings require confirmation in further studies. Until data are available to settle this larger issue, our findings regarding factors associated with bilateral disease may be useful in guiding clinical decisions regarding the extent of surgery in children with thyroid cancer.

In summary, in this analysis of pediatric patients with differentiated thyroid cancer, we observed no independent association between the risk of bilateral disease and any preoperative characteristic, except a borderline association with radiation exposure. Thus, in patients with a high preoperative likelihood of thyroid cancer, the question of whether total thyroidectomy is needed due to the risk of bilateral disease is unlikely to be informed by patient or nodule characteristics, with the possible exception of historical radiation exposure. Clinically evident lymph node involvement (cN1) is not independently associated with bilateral disease but may prompt total thyroidectomy to facilitate postoperative radioiodine therapy. For children diagnosed with thyroid cancer after lobectomy, a histological finding of multifocal disease or the entire lobe occupied by disease is associated with a markedly increased risk of residual disease in the contralateral lobe that should prompt consideration of completion thyroidectomy. Further studies are needed to determine whether tumor multifocality or bilateral disease is associated with adverse outcomes in children with thyroid cancer.

Glossary

Abbreviations

FNA

fine-needle aspiration

FTC

follicular thyroid carcinoma

OR

odds ratio

PTC

papillary thyroid carcinoma

Additional Information

Disclosures: The authors have nothing to disclose.

Data Availability

Datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.