Thyroid Ultrasound Screening in Childhood Cancer Survivors following Radiotherapy

Childhood cancer survivors (CCS) are at significant risk for developing late-onset endocrine complications from radiotherapeutic exposures, with prevalence of endocrinopathies as high as 50% [1, 2]. Among late effects, the development of thyroid disease, thyroid nodules, and differentiated thyroid cancer (DTC) are common [1, 3‒7]. The estimated lifetime risk for developing DTC among cancer survivors who received radiotherapy (RT) varies greatly, with standardized incidence ratios for radiation-associated DTC between 5- and 69-fold higher compared to controls [8‒11]. Previous studies demonstrate a linear dose-response relationship between radiation dosage and incidence of DTC, with the risk peaking at radiation levels between 10 and 30 Gy [9]. Younger age at the time of RT along with female sex are additional key determinants influencing the latency and likelihood of developing secondary DTC [10, 12, 13]. As the risk of DTC persists throughout adulthood, active surveillance and long-term follow-up have been advocated for survivors treated with radiation during childhood.

An ongoing debate remains on the optimal screening method for radiation-induced DTC in CCS. Current Children’s Oncology Group Follow-Up Guidelines recommend annual neck palpation and conservative use of ultrasound (US) for palpable nodules [14]; endocrine professional societies advocate for more precise monitoring with thyroid US, regardless of palpation results [15]; and guidelines from the International Late Effects of Childhood Cancer Guideline Harmonization Group recommend that providers decide between palpation and US on a patient-by-patient basis [11]. Those in favor of US surveillance refer to evidence showing that the use of US is associated with identification of DTC at an earlier stage prior to metastasis, with the potential benefit of achieving remission with less extensive surgical and medical therapy and reduced risk of complications [11, 16, 17]. Those against US surveillance suggest that it may lead to an increased risk of false-positive results, inducing anxiety among survivors and medical providers and resulting in unnecessary interventions [18, 19].

The lack of consensus revolving around best practices for the management of CCS at risk for developing secondary DTC prompted surveillance of CCS using thyroid US imaging at the Children’s Hospital of Philadelphia (CHOP). In this study, we investigated CCS of primary malignancies treated with RT and followed at our institution with thyroid US surveillance to assess the potential clinical benefit of thyroid US to detect radiation-induced DTC. We aimed to examine the outcome of using thyroid US to surveil for radiation-induced DTC in CCS and to determine whether there is an association between time at initiation of thyroid US surveillance from RT and extent of metastasis using the American Thyroid Association (ATA) pediatric thyroid cancer risk levels [20]. Enhanced understanding of the clinical course of radiation-induced DTC may optimize survivorship surveillance and management of the increasing population of pediatric cancer survivors.

An IRB-approved retrospective study was conducted of CCS with history of external beam RT for treatment of a non-thyroid cancer, bone marrow failure syndrome requiring total body irradiation (TBI) for allogeneic hematopoietic stem cell transplantation, or high-risk neuroblastoma survivors with previous treatment history of ¹³¹I-Metaiodobenzylguianidine (¹³¹I-MIBG) or TBI followed at CHOP between January 2002 and December 2021. CCS diagnosed with their primary malignancy <21 years old were included to parallel the CCS Study [21]. Patient information pertaining to primary malignancy and treatment, thyroid sonographic, cytologic, and pathologic features, and endocrine-related outcomes was collected. Fine-needle aspiration (FNA) cytology was classified according to the Bethesda System for Reporting Thyroid Cytopathology (TBSRTC) [22]. Primary tumor (T), regional lymph nodes (N), and distant metastasis (M) were staged using the 7th or 8th edition of the American Joint Committee on Cancer (AJCC) classification for DTC [23, 24]. Thyroid cancer risk stratification was adopted from the 2015 ATA Management Guidelines for Children with Thyroid Nodules and DTC [20]. Thyroid disease status at 1-year post-treatment was adopted from Nies et al. [25] Somatic driver oncogenic alterations were detected using CHOP’s Comprehensive Solid Tumor Panel (CSTP), a next-generation sequencing panel that encompasses 238 cancer genes and more than 600 fusions [26, 27].

Continuous variables were summarized using mean ± standard deviation or median (interquartile range [IQR] = 1st–3rd quartiles) depending on the data distribution. Categorical variables were summarized using frequency and percentage. For thyroid nodule outcomes, cutpoints for CCS RT age that are associated with varying risks of nodule presentation were explored. R package cutpointr was used, with nodule presentation as the outcome and age as the predictor. Cutpoints were selected to maximize the sum of sensitivity and specificity. Patient outcomes pertaining to thyroid nodules were then summarized by the age categories defined by the selected cutpoints. When the date of RT could not be confirmed, the date of bone marrow transplantation with TBI conditioning for leukemia patients and the date of primary diagnosis for non-leukemia patients were utilized. Categorical variables were compared using two-tailed Fisher exact test, and continuous variables were compared using Kruskal-Wallis test (nonparametric) or ANOVA method (parametric) with subsequent post hoc testing for pairwise group comparisons. Risk factors for thyroid nodule presentation were also evaluated using logistic regression, with the magnitude of association summarized by odds ratios (ORs) and 95% confidence intervals. Univariate logistic regression was first constructed for each risk factor; those that demonstrated some associations (univariate p < 0.1) were selected into the multivariate model. For the outcome of DTC, summary statistics were provided for the overall cohort; stratified analyses by age category were not performed due to small sample size of DTC. p values ≤0.05 were considered statistically significant. All analyses were performed in R 4.1.0 and R Studio 1.4.1717 [28‒31].

To identify our cohort, a list of 640 patients seen within the Division of Oncology and with a thyroid US ordered was generated by querying the hospital electronic medical record. An additional list comprising 237 patients was retrieved from CHOP’s Endocrine Late Effects after Cancer Therapy (ELECT) Program. The lists were cross-checked, confirming 20 additional patients for potential study inclusion. Of the 660 patients, 354 were excluded secondary to having no primary non-thyroid childhood malignancy (n = 234), no history of RT (n = 64), or no surveillance thyroid US despite history of RT for a primary malignancy (n = 56). Of these 56 patients, 37 (66%) had an US ordered but not performed, and 19 (34%) had an US performed for indications other than surveillance, including nodule(s) visualized on previous imaging or palpation (n = 14), thyroid-related symptoms (n = 3), or Hodgkin’s surveillance (n = 2). A cohort of 306 CCS (172 males and 134 females) who underwent thyroid US surveillance following exposure to RT, TBI, and/or MIBG was confirmed (shown in Fig. 1).

Demographic and clinical characteristics of the CCS cohort are described in Table 1. CCS were diagnosed with their primary malignancy at a median age of 4.8 years (IQR = 2.6–8.9). The most common CCS diagnoses were leukemia (32%), CNS tumor (26%), and neuroblastoma (18%). Patients received TBI (45%) and/or RT to craniospinal (44%), chest (11%), neck (6%), and other regions, with 53% of patients receiving a radiation dose ≥5 and <30 Gy and 43% of patients receiving a radiation dose ≥30 Gy. Patients received a median radiation dose of 23 Gy (IQR = 12–51). Of the patients with high-risk neuroblastoma, 27% (15/55) received ¹³¹I-MIBG. Median oncology follow-up following primary diagnosis was 13.9 years (range = 2.5–24.7; IQR = 10.1–18.1). Several endocrinopathies were diagnosed: 58% (176/306) presented with hypothyroidism, 56% (170/306) with growth hormone deficiency, 29% (89/306) with gonadal hormone deficiency, and 13% (41/306) with adrenal insufficiency.

Thyroid US surveillance data of the CCS cohort are presented in Table 2. Age cutpoint analyses identified cutpoints 3 and 10; hence, ≤3, >3 to ≤10, and >10 years were used. Thyroid US surveillance was initiated at a median interval of 9.1 years (range = 0.3–23.8; IQR = 6.0–12.0) following RT. Thyroid nodule(s) were detected in 49% (150/306) of patients, with nodule(s) identified on initial US screening in 71% (107/150) of patients and subsequent thyroid US monitoring in 29% (43/150) of patients. The median number of USs performed for the initial detection of nodule(s) was 1 (range = 1–8). Median ages at nodule discovery were 15.7 years (range = 6.0–21.0) for patients ≤3, 17.3 years (range = 10.2–27.4) for patients >3 to ≤10, and 22.1 years (range = 14.8–29.7) for patients >10 years at the time of RT (p < 0.001). Of the patients with nodule(s), 61% (91/150) presented multinodular disease (2+ nodules), and 39% (59/150) presented a solitary nodule. Interval from RT to the first US (p < 0.001), interval from RT to the initial presentation of nodule(s) (p < 0.001), and incidence of nodule(s) (p = 0.002) decreased as age at the time of RT increased. Nodule(s) were detected in 68% (41/60) of patients ≤3, 46% (84/181) of patients >3 to ≤10, and 39% (25/65) of patients >10 years at the time of RT. In multivariate logistic regression, RT for patients ≤3 years old conferred a 2.5-fold increased risk for nodule presentation compared to RT for patients >10 years old (OR = 2.87 [2.06–3.68] p = 0.011). Female sex (OR = 1.62 [1.13–2.12] p = 0.054) and greater interval between RT and first US (OR = 1.10 [1.04–1.16] for every 1-year increase p = 0.001) were additional independent risk factors for nodule(s).

In univariate and multivariate logistic regressions, radiation dosages <30 Gy were not a significant risk factor for the development of a thyroid nodule (OR = 1.51 [1.01–2.01] p = 0.104). There was no statistical difference in the median radiation dose between nodule-positive (23 Gy; IQR = 13–50) and nodule-negative patients (25 Gy; IQR = 12–54; p = 0.653). More than half (59%; 84/142) of patients presenting with nodule(s) were treated with a radiation dose ≥5 and <30 Gy; 40% (57/142) of nodule-positive patients were treated with a radiation dose ≥30 Gy. Of the nodule-positive patients ≤3 years at the time of RT, 63% (26/41) were treated with a radiation dose ≥5 and <30 Gy. Radiation doses for 9 patients were unknown and not reported.

Of the patients with nodule(s), 37% (55/150) underwent FNA secondary to suspicious feature(s) on US (n = 52) or to confirm the absence of malignancy for nodules with benign US characteristics (n = 3). Approximately 63% (95/150) of patients presenting with nodule(s) did not undergo FNA at CHOP as they transitioned to adult care or were lost to follow-up (n = 44), had low-risk features on US (n = 41), were deceased (n = 5), or underwent surgery secondary to multinodular goiter (n = 4) or US features highly suggestive of malignancy (n = 1; shown in Fig. 2).

Forty-four (44/306; 14%) patients pursued thyroidectomy, and 64% (28/44) were diagnosed with thyroid malignancy on histology. Of the patients diagnosed with DTC, 64% (18/28) had thyroid cytology suspicious for (TBSRTC V) or consistent with malignancy (TBSRTC VI). The 16 patients in our cohort who ultimately presented with benign histology pursued surgical intervention for indeterminate cytology (TBSRTC III-IV; n = 11), to discontinue surveillance (n = 4), or for unsatisfactory cytology (TBSRTC I; n = 1). The median time between RT exposure and onset of DTC was 11.4 years (range = 5.4–17.9; IQR = 9.0–14.1), and the overall rate of DTC in our cohort was 9% (28/306). The incidence of DTC was 11% (26/241) in patients ≤10 years compared to 3% (2/65) in patients >10 years at the time of RT exposure. There was no difference in the incidence of DTC between males (8%; 14/172) and females (10%; 14/134).

Clinicopathologic features of the 28 patients with DTC are presented in Table 3. Of these patients, 93% (26/28) were diagnosed with papillary thyroid carcinoma (PTC), with classic variant of PTC as the most common histologic subtype. The median radiation dose for patients with DTC was 23 Gy (range = 10–79; IQR = 12–54). Approximately 61% (17/28) of DTC patients were treated with a radiation dose ≥5 and <30 Gy compared to 39% (11/28) of DTC patients treated with a radiation dose ≥30 Gy. Based on the ATA pediatric DTC risk stratification, 68% (19/28) of patients presented with low-risk, 7% (2/28) with intermediate-risk, and 25% (7/28) with high-risk disease, with 57% (4/7) of the high-risk patients demonstrating pulmonary metastasis. Of the 9 (32%) patients with ATA intermediate- or high-risk disease, 8 (89%) were ≤10 years and 1 (11%) was >10 years at the time of exposure to radiation. The primary cancers in these patients were CNS tumors (n = 4), leukemia (n = 3), or neuroblastoma (n = 2). These patients had a history of cranial/craniospinal radiation (n = 5), TBI (n = 4), abdominal/pelvic radiation (n = 2), and/or MIBG therapy (n = 1). The 2 patients treated with abdominal/pelvic radiation also received radiation with TBI (n = 1) and MIBG therapy (n = 1), respectively.

Of the 4 patients with pulmonary metastasis, 100% received craniospinal radiation for diagnosis of CNS tumor (n = 3) or leukemia (n = 1). The youngest patient to present with pulmonary metastasis was 9 years old, and the minimum interval between RT exposure and diagnosis of metastatic DTC was 5.7 years. RAI (¹³¹I) therapy was administered to 3 of these patients. One CNS tumor survivor was placed on an NTRK-oncogene-specific inhibitory therapy secondary to presenting hypoxia that was associated with extensive lung disease burden from PTC. Of the 13 DTC patients who underwent somatic molecular testing using CHOP’s CSTP, 11 had somatic pathogenic fusions associated with high risk for invasive DTC (shown in Fig. 2). Three of the 4 patients with pulmonary metastasis underwent somatic molecular testing and were found to have a somatic fusion oncogene: SPECC1L::RET fusion (n = 2) or TPR::NTRK1 fusion (n = 1).

We analyzed 306 CCS with a history of radiation exposure who underwent definitive surveillance thyroid US imaging during the 20-year study period. The integration of thyroid US imaging in survivorship care detected thyroid nodule(s) in 49% (150/306) of patients in our CCS cohort, with 18% (55/306) of patients undergoing FNA and 14% (44/306) of patients pursuing surgical resection (shown in Fig. 2). Our results and prior studies indicate that selective use of FNA in CSS has high sensitivity and specificity for detecting malignant thyroid nodules and that the application of US in clinical practice for CCS is not associated with an increased rate of referrals for additional interventions [32]. The incidence of DTC was 9% (28/306) in our entire cohort and 19% (28/150) in nodule-positive CCS. The rate of radiation-induced DTC is similar to the rate of malignancy in sporadic pediatric nodules (19% vs. 22–34%) [20, 32, 33] and in accordance with several studies comprising smaller CCS cohorts that evaluated the screening utility of thyroid US (9% vs. 6–11%) [34‒37]. Our rate of thyroid malignancy is higher than rates reported in large CCS epidemiological studies [8, 10, 38, 39].

The majority (68%; 19/28) of CCS harboring radiation-induced DTC in our cohort were classified as ATA pediatric low-risk for persistent disease, with half (50%; 14/28) presenting no evidence of regional lymph node metastasis and the majority (86%; 24/28) presenting no evidence of distant metastasis. However, one-third (32%; 9/28) of CCS with DTC demonstrated metastatic disease. Sites of metastases included the lymph nodes (N1a/N1b; n = 8), lungs (n = 4), and/or mediastinum (n = 1; shown in Table 3). At the extreme, 1 CNS tumor survivor presented with hypoxia and advanced PTC-associated pulmonary metastasis and was placed on NTRK-specific targeted systemic therapy to decrease tumor burden prior to administering RAI. Upon review of preoperative USs for CCS with ATA intermediate- or high-risk DTC, 55% (5/9) of primary tumors would not have been detected by physical examination due to no displacement to the anterior margin of the thyroid lobe. Of these patients, 1 demonstrated reduced range of neck motion secondary to radiation-induced cervical spine disease and was ultimately diagnosed with AJCC N1b disease. If thyroid US was not utilized in the clinic, it is probable that these patients would have had an even greater delay in diagnosis of secondary DTC, likely increasing the risk of disease burden.

Our study suggests that irradiation of CCS at younger ages may influence the development of DTC, a phenomenon that is well-established in the literature [12, 40‒42]. Moreover, the observation that 39% (11/28) of CCS demonstrating DTC received radiation doses ≥30 Gy suggests that the risk for developing DTC may persist even at high cumulative doses. While the median interval between RT and the first US for CCS was 9.1 years (range = 0.3–23.8), we concurrently observed CCS developing DTC following a shorter latency from RT exposure. Importantly, the minimum latency for developing PTC-associated pulmonary metastasis was found to be 5.7 years post-RT. The distribution of ATA intermediate- and high-risk DTC in CCS ≤10 years old at the time of RT suggests that delayed initiation (>5 years) of thyroid US surveillance post-RT exposure for younger patients (≤10 years) may be associated with an increased risk for patients being diagnosed with lateral neck (AJCC N1b) and distant (AJCC M1) metastases. In a population that bears significant risk for secondary neoplasms following RT, we believe US surveillance may provide an opportunity to improve clinical outcomes of CCS by detecting abnormal thyroid nodules and DTC at an earlier state of metastasis. While there is mixed data regarding whether radiation-induced DTC is associated with an increased risk of metastasis [43, 44] or not [45, 46], outcomes from our CCS cohort support the incorporation of thyroid US surveillance for CCS exposed to radiation at younger ages, regardless of cumulative RT dose [15].

Based on our findings, early thyroid monitoring with US imaging may bestow an opportunity to diagnose radiation-induced DTC at an earlier AJCC TNM stage, reducing the potential need for extensive surgery and systemic therapy [47]. At a high-volume pediatric thyroid program, the incorporation of thyroid US surveillance does not appear to increase the rate of unnecessary referral for FNA as only 18% (55/306) of patients were referred for biopsy. The varying incidence of thyroid nodules (68%; 46%; 39%) across CCS age cohorts suggests an age-stratified approach to initiation of thyroid US surveillance. The lack of a difference in the median radiation dose between nodule-negative patients (25 Gy; IQR = 12–54), nodule-positive patients (23 Gy; IQR = 13–50), and DTC patients (23 Gy; IQR = 12–54) suggests age at the time of RT to be a stronger risk factor compared to cumulative RT dosage for radiation-induced thyroid disease. Lastly, the wide interval range (5.4–17.9 years) for CCS developing DTC following RT accentuates the importance of long-term follow-up care.

This cohort study is limited by its single-center sample size, lack of a control group for outcome comparisons, and the evolving nature of CCS management post-RT over the past two decades. Consequently, the total number of CCS treated with RT and followed at CHOP over the study period, total radiation dosage for 9 patients, and the dose delivered to the thyroid gland could not be determined. The lack of information on CCS who transitioned to adulthood or were lost to follow-up could have led to underestimation in the incidence of nodules and DTC, particularly for CCS treated with RT >10 years old. Due to the retrospective nature of this study, the latency for developing DTC post-RT exposure cannot be fully confirmed. Though 67% (8/12) of CCS with oncogenic drivers demonstrated invasive DTC, further investigation of the molecular landscape and tumor microenvironment of radiation-induced DTC is required to draw further conclusions [48, 49]. Despite these limitations, this study employs the largest cohort of CCS to date and presents valuable information that contributes to the ongoing discussion regarding DTC surveillance guidelines in CCS.

To improve the management of CCS, we have established a Thyroid US Surveillance Clinic to perform point-of-care thyroid USs for CCS by trained clinicians. In the clinic, we are prospectively screening CCS to further (1) define the latency to developing DTC following RT exposure to confirm when US surveillance should be initiated and (2) identify risk factors to confirm the cohort of CCS who would have the greatest benefit from US surveillance.

Thyroid US surveillance appears beneficial in CCS exposed to RT at younger ages to detect nodules with the benefit of diagnosing DTC at an earlier AJCC TNM stage, prior to metastasis to the lateral neck or lungs. Prospective studies that aim to further define the latency between RT exposure and development of nodules and DTC are warranted to refine thyroid surveillance guidelines for CCS at high risk. Based on our data and the lack of prospective data in the literature, we plan to prospectively determine the advantage of thyroid US surveillance in CCS with exposure to RT at younger ages (≤10 years) compared to CCS with exposure at older ages (>10 years) to mitigate unnecessary interventions while avoiding delayed diagnosis.

This study involving human subjects was reviewed and approved by the Children’s Hospital of Philadelphia Institutional Review Board (CHOP IRB #17-014224). Written informed consent from the participant and/or the participant’s legal guardian was not required per CHOP IRB; a waiver of consent/parental permission has been approved per 45 CRF 46.116(d).

Research was conducted in the absence of commercial or financial conflicts.

This work was supported by the National Cancer Institute (K07 CA166177; SMM), Sanford Chair in Pediatric Oncology (SMM), and Thyroid Center Frontier Program (AJB).

Conception and design: S.M.M. and A.J.B. Data collection, analysis, and assembly: J.A.B. and S.H. Data interpretation: J.A.B., S.H., S.M.M., A.J.B., Y.L., A.I., K.K., N.S.A., and J.P.G. Administrative support: A.I., T.P., and L.S. Manuscript draft: J.A.B., S.H., A.J.B., and S.M.M. Manuscript editing and final approval for all aspects of the work: all authors.

All data generated and analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.