Controversies in the Management of Low-Risk Differentiated Thyroid Cancer

Abstract

Controversy exists over optimal management of low-risk differentiated thyroid cancer. This controversy occurs in all aspects of management, including surgery, use of radioactive iodine for remnant ablation, thyroid hormone supplementation, and long-term surveillance. Limited and conflicting data, treatment paradigm shifts, and differences in physician perceptions contribute to the controversy. This lack of physician consensus results in wide variation in patient care, with some patients at risk for over- or undertreatment. To reduce patient harm and unnecessary worry, there is a need to design and implement studies to address current knowledge gaps.

In this review, we evaluated the current management of low-risk differentiated thyroid cancer and identified areas of controversy.

Essential Points

• Limited and conflicting data secondary to low event rates and small sample sizes contribute to the controversy in the management of low-risk differentiated thyroid cancer

• Differences in physician specialty, experience, and general attitude toward intensive treatment influence treatment recommendations

• Changes in clinical guidelines over time compound the controversy

• Treatment decisions are interrelated with decisions on intensity of surgery, use of radioactive iodine for remnant ablation, thyroid hormone supplementation, and long-term follow-up closely linked

• The manifestation of the controversy is variation in care, with patients at risk for over- and undertreatment

• Treatments are not without risks, and benefits-risks need to be balanced and tailored to the patient

• To reduce controversy in the management of low-risk differentiated thyroid cancer, it is necessary to design studies to address knowledge gaps and to disseminate the findings broadly

Controversy exists across all phases of management of low-risk differentiated thyroid cancer (DTC), including surgery, radioactive iodine (RAI) for remnant ablation, thyroid hormone supplementation, and long-term surveillance. The etiology of the controversy is likely multifactorial and related to rising incidence, indolent disease course, treatment paradigm shifts, and, in some instances, limited or conflicting data. The manifestation of this controversy is wide variation in care, which has implications for patients, providers, and policy makers.

The controversy in management of low-risk DTC starts with the controversy in defining “low risk.” Approximately 90% of all patients with DTC, including patients with papillary thyroid cancer (PTC) and follicular and Hurthle cell cancer, have low-risk disease if the definition is based on disease-specific survival (DSS). In these “low-risk” patients, the 10-year DSS is 95% to 100% (1, 2). Most of these patients would be classified as American Joint Committee on Cancer (AJCC) Tumor, Node, Metastasis (TNM) stage I or II. The AJCC TNM staging system takes into consideration patient age, tumor size, cervical lymph node status, and presence/absence of distant metastases when predicting risk of death from DTC (3). Even though the relationship between patient age and death from thyroid cancer is linear (2, 4), as shown in Table 1, the AJCC TNM staging system categorizes patients with DTC based on age 45 years in the seventh edition and age 55 years in the eighth edition (3, 5). However, this well-accepted staging system is focused solely on predicting survival and is not intended to predict recurrence risk, which is a more common occurrence for DTC patients. The overall risk of recurrence from DTC is 20% to 30%, with the range varying between 1% and 70% based on tumor characteristics (6, 7). As shown in Table 2, the American Thyroid Association (ATA) Risk Stratification System defines “low risk” based on the likelihood of recurrence (8, 9). This risk stratification system incorporates pathological variables and, in some scenarios, mutational status. It is recognized that risk of recurrence is a continuum, with recurrence occurring in 1% to 5% of patients with low-risk DTC (8).

Table 1.

Defining Low Risk Based on Survival

AJCC TNM Seventh Edition	AJCC TNM Eighth Edition
<45 years and stage I	<55 years and stage I
Any T, any N, M0	Any T, any N, M0
<45 years and stage II	<55 years and stage II
Any T, any N, M1	Any T, any N, M1
≥45 years and stage I	≥55 years and stage I
T ≤2 cm, N0, M0	T ≤2 cm, N0/NX, M0
	T >2 cm but ≤4 cm, N0/NX, M0
≥45 years and stage II	≥55 years and stage II
T >2 cm but ≤4 cm, N0, M0	T ≤2 cm, N1, M0
	T >2 cm but ≤4 cm, N1, M0
	T >4 cm or any T and gross ETE invading only strap muscle, any N, M0

Data from Edge et al. (3, 5).

Abbreviations: M0, no distant metastases; M1, distant metastases; N0, no evidence of regional lymph node metastasis; N1, metastases to cervical/upper mediastinal lymph nodes; NX, unknown status of regional lymph node metastasis; T, tumor size.

Table 2.

Defining Low Risk Based on Recurrence

ATA Risk Stratification
PTC (with all of the following):
	No local or distant metastases
	All macroscopic tumor has been resected
	No tumor invasion of loco-regional tissues or structures
	The tumor does not have aggressive histology (e.g., tall cell, hobnail variant, columnar cell carcinoma)
	If I-131 is given, there are no RAI-avid metastatic foci outside the thyroid bed on the first post-treatment whole-body RAI scan
	No vascular invasion
	Clinical N0 or ≤five pathologic N1 micrometastases (<0.2 cm in largest dimension)
Intrathyroidal, encapsulated follicular variant of PTC
Intrathyroidal, well-differentiated follicular thyroid cancer with capsular invasion and no or minimal (<four foci) vascular invasion
Intrathyroidal, papillary microcarcinoma, unifocal, or multifocal, including BRAF V600E mutated (if known)

Data from Haugen et al. (8).

Abbreviations: I-131, iodine-131; N0, no evidence of regional lymph node metastasis; N1, metastases to cervical/upper mediastinal lymph nodes.

One theory is that some low-risk disease may represent the early stage of high-risk disease. However, the slow progression of DTC in general, the low mortality rates, and the high incidence of small thyroid cancers that are discovered on autopsy make this less probable (1, 10, 11). In addition, certain variants of PTC and follicular thyroid cancer more commonly have low-risk phenotypes. For example, classic PTC, minimally invasive follicular thyroid cancer, and the entity recently reclassified as a noncancerous lesion, noninvasive follicular variant of PTC, typically have a more indolent disease course (12–14). In addition, specific genomic alterations are more common in high- vs low-risk disease (15). For example, coexisting BRAF V600E and TERT C228T mutations generally lead to a high-risk phenotype; therefore, this combination of mutations would likely be uncommon in phenotypically low-risk disease (16). Although there is a continuum in terms of the clinical presentation of DTC, typically, low-risk DTC has a distinct phenotype from high-risk disease and likely does not represent early detection of the latent phase of aggressive disease.

In the United States, the incidence of low-risk DTC has increased threefold in the last 30 years (11, 17). The rise in incidence has been most rapid for tumors ≤2 cm (11, 18). A similar rise in incidence has been observed worldwide, with the greatest increase seen in South Korea (19, 20). Although the majority of this rise in incidence is attributed to overdiagnosis secondary to excess imaging, in the United States between 1974 and 2013, there was an increase in both the incidence and mortality rate for advanced-stage PTC (21). This suggests that the rise in incidence is multifactorial, with a true increase in occurrence contributing to the epidemic. Irrespective of etiology, the worldwide rise in incidence of low-risk DTC emphasizes the importance of defining optimal treatment.

As shown in Tables 3 and 4, over the past 10 years, the ATA guideline recommendations for optimal treatment of DTC have changed (8, 9, 22, 25–27). Over time, the guidelines have increasingly emphasized more fastidious risk stratification to inform potentially less aggressive surgery, use of RAI, and use of thyroid hormone suppression. Despite this pendulum shift in clinical guidelines, in practice, there remains tremendous variation in the management of low-risk DTC (23, 24). As shown in Table 5, this practice variation can lead to two clinically identical patients receiving markedly different treatments based on intensity. In a survey of 944 physicians involved in thyroid cancer care, Haymart et al. (24) calculated the ratio of observed variation in management to hypothetical maximum variation with maximum measured variation as 1.00. For low-risk DTC, there is variation in extent of surgery [variation, 0.91; 95% confidence interval (CI), 0.88 to 0.94], the role of RAI (variation, 0.91; 95% CI, 0.89 to 0.93), and the role of suppressive doses of thyroid hormone replacement (variation, 1.00; 95% CI, 0.99 to 1.00). In addition, variation existed in all aspects of long-term management (24). Results from physician survey studies led by Haymart et al. (24, 28–30) imply that the wide hospital-level variation in the treatment of thyroid cancer is not explained by differences in patient case-mix but is more likely related to differences in physician decision-making. In turn, this is likely related to physician specialty, experience, and general attitude toward intensive treatment (24, 28–30). With marked treatment variation, some low-risk patients may be at risk for overtreatment and potential harm from treatment of indolent disease, whereas others may be at risk for undertreatment of clinically significant disease. This variation in care further emphasizes the importance of understanding the ongoing controversy in the management of low-risk DTC.

Table 3.

ATA Guidelines on Low-Risk DTC Management Over the Past 10 years

	2006	2009	2015
Surgery
Total thyroidectomy vs lobectomy	Recommend total thyroidectomy if DTC >1–1.5 cm, contralateral thyroid nodules, regional or distant metastases, history of head/neck radiation, or first-degree family member with thyroid cancer; age > 45 also may result in total thyroidectomy	Recommend total thyroidectomy if DTC >1 cm	Recommend either total thyroidectomy or lobectomy for cancer >1 cm and <4 cm without ETE and nodal metastases
	Lobectomy may be sufficient if small (<1 cm), low-risk, isolated intrathyroidal PTC without nodal metastases	Lobectomy may be sufficient for small (<1 cm), low-risk, unifocal intrathyroidal PTC if no history of head or neck radiation or clinically involved nodal metastases	Recommend lobectomy for cancers <1 cm without ETE or nodal metastases, unless indications to remove contralateral lobe
Prophylactic central neck dissection	Routine pCLND should be considered for patients with PTC and suspected Hurthle cell cancer	Routine pCLND may be performed in patients with PTC, especially if T3 or T4 tumors	pCLND should be considered for PTC if tumors are T3 or T4, if lateral neck lymph nodes are clinically involved, or if the information will help plan further treatment
RAI ablation
When to treat	Recommended for all patients with stage II disease <45 years, most patients with stage II disease >45 years	Remnant ablation is not recommended if unifocal or multifocal tumor <1 cm without other high-risk features	RAI remnant ablation is not routinely recommended for unifocal PTmC, multifocal PTmC, and other low-risk DTC; however, specific features of the patient that affect recurrence risk, disease follow-up implications, and patient preferences should be considered
	Recommended for select stage I disease and if multifocal disease, nodal metastases, extrathyroidal or vascular invasion, and/or more aggressive histologies	Selective use for other low-risk patients when age, tumor size, lymph node status, and individual histology predict an intermediate to high risk of recurrence or death
Preparation	LT4 withdrawal or rhTSH stimulation	LT4 withdrawal or rhTSH stimulation	rhTSH stimulation is an acceptable alternative to withdrawal
Dose for remnant ablation	30–100 mCi	30–100 mCi	30 mCi is favored over higher doses
Goal TSH
Initial goal	0.1–0.5 mIU/L for low-risk patients	0.1–0.5 mIU/L for low-risk patients	0.5–2.0 mIU/L unless there is a low-level serum Tg, in which case goal is 0.1–0.5 mIU/L; after total thyroidectomy with RAI, Tg ≥0.2 ng/mL is considered to be low level, whereas the Tg cutoff is less clear if total thyroidectomy and no RAI or if lobectomy
Long-term goal	0.3–2.0 mIU/L in patients free of disease, especially if low risk of recurrence	0.3–2.0 mIU/L in patients free of disease, especially if low risk of recurrence	0.5–2.0 mIU/L in patients with an excellent or indeterminate response to therapy, especially those at low risk for recurrence

Data from Haugen et al. (8) and Cooper et al. (9, 22). Staging is based on AJCC TNM 7th edition.

Abbreviations: Tg, thyroglobulin.

Table 4.

ATA Compared to NCCN Guidelines

	2006–2007	2009–2010	2015–Present
ATA
Surgery			Can consider active surveillance for PTmC with low-risk features
	Lobectomy for tumors <1 cm and age < 45	Lobectomy for tumors <1 cm	Lobectomy for T1-2 tumors
	TT for tumors >1 cm	TT for tumors >1 cm	TT for T3-4 tumors
pCLND	Routine for all PTC	May be considered for T1-2	Should be considered for T3-4 tumors
		Recommended for T3-4
NCCN
Surgery	Lobectomy PTmC	Lobectomy PTmC	Lobectomy PTmC
	Lobectomy (or TT) age 15–45 or tumor >1 cm and <4 cm	Lobectomy (or TT) age 15–45 or tumor >1 cm and <4 cm	Lobectomy (or TT) tumor <4 cm
	TT age < 15 or > 45 or tumor >4 cm	TT age < 15 or > 45 or tumor >4 cm	TT for tumor >4 cm
pCLND	No recommendation	Consider for age < 15 or > 45 or tumor >4 cm	Consider for tumor >4 cm

Data from Haugen et al. (8), Cooper et al. (9, 22), Haddad et al. (25), Tuttle et al. (26), and Sherman et al. (27).

Abbreviations: ATA, American Thyroid Association; NCCN, National Comprehensive Cancer Network; TT, total thyroidectomy.

Table 5.

Low- vs High-Intensity Treatment of Low-Risk Thyroid Cancer

	Low-Intensity Treatment	High-Intensity Treatment
Surgery	Lobectomy	Total thyroidectomy with prophylactic central neck dissection
RAI	None	High-dose RAI with thyroid hormone withdrawal
Suppressive doses of thyroid hormone	None	TSH suppression
Benefits	Fewer risks to patients, including lower risk of recurrent laryngeal nerve injury, hypoparathyroidism, sialadenitis, osteoporosis, arrhythmia, etc.	Easier long-term follow-up with tumor markers and neck ultrasound
	Possible avoidance of thyroid hormone supplementation	Less uncertainty about response to therapy and cancer-free status
	Lower costs

Example above: 1.5-cm PTC with no worrisome lymph nodes on ultrasound. The same patient with low-risk PTC could receive markedly different treatment intensity. The low- and high-intensity treatments reflect variation in practice patterns, not guideline recommendations.

In this review, controversies in the management of low-risk DTC are evaluated across four domains: surgery, RAI, thyroid hormone supplementation, and long-term surveillance. The literature search emphasizes contemporary studies, with a focus on publications from 2006 to 2016. However, in certain scenarios, pivotal historical references were included.

Controversies in Management

Surgery

Surveillance vs surgery

Historically, the first-line therapy for DTC has been surgery. However, surgery is associated with the risk of complications and the need for levothyroxine treatment in 22% of patients following thyroid lobectomy, or replacement in 100% of patients following near-total or total thyroidectomy (31). In addition to the generic risks of anesthesia, including cardiovascular and pulmonary complications, morbidity associated with thyroid surgery can include temporary and permanent alterations in voice associated with injury to the recurrent and/or superior laryngeal nerves, hematoma, need for reoperation, and infection. For near-total and total thyroidectomy, the risks also include hypoparathyroidism (temporary and permanent) and potentially loss of airway in the case of bilateral recurrent laryngeal nerve injury. These risks must be balanced against the potential benefit associated with resection of the cancer. Although there is general consensus that benefit outweighs risk for surgery performed for clinically important DTC >1 cm (8), there is a growing body of evidence suggesting that surveillance potentially can serve as a safe alternative to surgery for at least papillary thyroid microcarcinoma (PTmC) (32–35). Indeed, the current ATA guidelines do not recommend fine needle aspiration (FNA) biopsy of nodules <1 cm, even when PTmC is suspected based on suspicious ultrasound characteristics, such as hypoechogenicity with either microcalcifications, taller-than-wide tumor dimensions, rim calcification with tumor extrusion, or irregular margins, assuming absence of extrathyroidal extension (ETE) or lymph node metastasis (8).

A prospective observational trial of PTmC was initiated in 1993 at Kuma Hospital, Kobe, Japan, and a similar trial started at the Cancer Institute Hospital in Tokyo in 1995 (29–34). The body of data assembled from these prospectively followed cohorts currently represents the largest published experiences and longest clinical follow-up whereby active surveillance was used and surgery was deferred for PTmC (32–35, 37). In these protocols, patients with thyroid nodules measuring ≤1 cm confirmed to be PTC by fine-needle aspiration biopsy were allowed to choose between two alternative treatment options: immediate surgery or active surveillance (34). Patients who had more concerning findings at diagnosis, including a tumor abutting the trachea, possibly invading the recurrent laryngeal nerve, cytology concerning for higher-grade malignancy, or lateral compartment lymph nodes highly suspicious for metastatic disease by ultrasound or confirmed by FNA, were excluded and offered immediate surgery (34). Between 1993 and 2001 at Kuma Hospital, Ito et al. (34) identified 732 patients, and 162 (22%) elected to be surveilled, whereas 571 elected to undergo immediate surgery. Unilateral, unifocal disease was treated with thyroid lobectomy and ipsilateral central compartment lymph node dissection, whereas bilateral multifocal disease or those with lateral compartment lymph node metastases were treated with total thyroidectomy, central compartment lymph node dissection, and ipsilateral lateral neck dissection (34). Follow-up ranged from 18 to 113 months (mean, 46.5 months) (34). The study authors observed that 15.3% of patients experienced an increase in the maximal diameter of their tumors by 2 mm within the first year of observation, and 27.5% experienced an increase in tumor size within 5 years of observation (34). Just 1.2% of patients developed new lateral neck lymph node metastases over the study period (34). Ultimately, 56 patients (35%) during the initial study timeframe went on to surgery as a result of enlargement of coexisting nodules or patient preference; however, only 9 of those 56 patients (5.6% of the total study cohort) were advised to have surgery due to clinical progression, defined as either an increase in tumor size to >1 cm (7 patients) or detection of lymph node metastases (2 patients), and none experienced a recurrence or thyroid cancer–related death (34).

Later studies from the Kuma Hospital migrated the criteria to 3 mm for what was considered an increase in tumor size, and longitudinal data analysis with at least 5 years of follow-up demonstrated that only 6.4% of patients experienced tumor enlargement at 5 years and 15.9% at 10 years (35). It is important to note that not all patients judged to have an increase in tumor size went on to have surgery, as 42% elected to continue observation despite an increase of ≥3 mm (35). The most contemporary update of this long-term observational study accrued 1235 patients from 1993 to 2011; only 4.6% experienced enlargement of their primary tumors by 3 mm, and 3.5% were advised to undergo surgery for “progression to clinical disease,” defined as tumor size that increased to ≥1.2 cm. Surgery was recommended for 1.5% of patients due to new nodal metastases (0.3% in the central compartment and 1.2% in the lateral compartment) over a mean follow-up period of 60 months (32). Among those patients followed for ≥10 years, 8.0% demonstrated tumor enlargement ≥3mm, 6.8% progressed to have tumors ≥1.2 cm, and 3.8% developed lymph node metastases (32). No patient demonstrated distant metastases or experienced a thyroid cancer–related death (32).

A series of 230 patients with 300 PTmCs from the Cancer Institute Hospital Tokyo was enrolled in an active surveillance program between 1995 and 2008 (33). With a mean follow-up of 5 years (range, 1 to 17 years), tumor enlargement (as measured by an increase of ≥3 mm from the original size of the primary tumor) was identified in 7.3% of lesions, and 1.3% of patients developed lateral nodal metastases (33). Of the 12 patients (5.2%) that ultimately went on to surgery for either tumor enlargement or lymph node metastases, none were reported to have had thyroid cancer recurrence during up to 12 years of follow-up (33). Taken together, these studies included 1465 patients with PTmC who were actively surveilled; 5.4% (67/1465) demonstrated tumor growth, and 1.5% (22/1465) developed lymph node metastases. There were no deaths related to thyroid cancer.

There are some situations where it may not be appropriate to enroll patients into an active surveillance program. In establishing an appropriate framework for active surveillance of PTmC, Brito et al. (38) outlined several interrelated domains (including the medical care team, tumor features, patient characteristics) with factors that can loosely stratify risk and guide the appropriateness of surveillance. If a medical care team is not experienced or equipped for an active surveillance program, then the patient would be better served with definitive surgical treatment (38). Characteristics of the patient, such as compliance or ability (through financial or medical insurance means) to adhere with the rigors of long-term follow-up, must be considered, as well as the patient’s preference or willingness to enter into an active surveillance program (39). Another patient attribute to factor into the decision-making process is patient age. Typically, the prognosis for elderly patients with PTC is compromised compared with that of younger patients (40, 41). However, Ito et al. (32) demonstrated that PTmC progressed less frequently in elderly patients aged ≥60 years (2.2%) compared with patients aged <40 years (5.9%) and 40 to 59 years (5.7%). The development of lymph node metastases among patients undergoing active surveillance also was lower among older patients, with only 0.4% of those aged ≥60 years developing metastatic disease during long-term active surveillance compared with 1.4% of patients 40 to 59 years and 5.3% of those aged <40 years (32). These findings are consistent with a study by Fukuoka et al. (36) of calcification patterns and PTmC vascularity by ultrasound during the course of active surveillance. Examining 384 patients enrolled in an active surveillance program for PTmC over a mean of 6.8 years, the authors demonstrated on ultrasound that tumor calcification patterns become more macroscopic and coarse over time (36). Coarse calcification patterns correlated with patient age, as the mean age of patients with no calcifications present was 50.7 years vs 60.1 years for patients with the most coalesced calcification patterns (36). With an increase in calcification, PTmC vascularity was observed to decrease, and both findings correlated strongly with disease nonprogression (36).

A significant concern in the current health care climate is cost and whether a given treatment strategy is cost-effective. Oda et al. (42) examined the cost of active surveillance compared with that of immediate surgery in the Kuma Hospital experience under the Japanese Health Care Insurance System. Factoring into their calculations the rate of conversion from active surveillance to delayed surgery, the authors found that the cost of immediate surgery was 4.1 times higher (928,094 yen/patient) than that of active surveillance over 10 years, with 8% of patients requiring delayed surgery (225,695 yen/patient) (42). Venkatesh et al. (43) more specifically examined the cost-effectiveness of active surveillance compared with thyroid lobectomy for PTmC. Using a reference case of a 40-year-old patient with unifocal PTmC and no other risk factors for aggressive PTmC, a Markov state-transition decision analysis model was created that cycled annually for a duration of 20 years (43). Thyroid lobectomy was more expensive ($13,866) than active surveillance ($5,724), assuming both groups were monitored annually with a physician office visit and ultrasound. However, thyroid lobectomy offered patients an increased effectiveness of 22.1 quality-adjusted life years (QALYs) compared with 20.3 QALYs for active surveillance (43). With 20 years of follow-up, the incremental cost-effectiveness ratio (ICER) for lobectomy of $4,437/QALY was well below the willingness-to-pay threshold of $100,000/QALY generally used for cost-effectiveness analyses (43). In the two-way sensitivity analysis, overall cost-effectiveness was highly dependent on the patient-specific decrease in quality of life associated with active surveillance and life expectancy following diagnosis (43).

The possible cost-effectiveness of immediate surgery compared with active surveillance needs to be interpreted in the context of extent of surgery for PTmC being limited to thyroid lobectomy. For unifocal PTmC, the ATA guidelines generally recommend surgery but also affirm active surveillance as an alternative strategy to immediate surgery (8). Should surgery be chosen, it is recommended to limit surgery to thyroid lobectomy unless other risk factors are present, including a history of cervico-facial radiation, multifocal and/or bilateral disease, and/or a family history of thyroid cancer (8). Despite these recommendations, most PTmC appears to be overtreated in the United States. In an analysis of 29,512 patients with a diagnosis of PTC in the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) database from 1988 to 2010, Wang et al. (44) found that 73.4% of patients with PTmC were treated with total thyroidectomy. The likelihood of total thyroidectomy increased with time, as 70.5% of patients underwent total thyroidectomy between 1988 and 1999, 72.4% between 2000 and 2006, and 77.7% between 2007 and 2010 (P < 0.001) (44).

Extent of surgery

For those patients who progress during active surveillance and those with DTC >1 cm at presentation, the current ATA guidelines recommend thyroid surgery (8). There has been considerable debate over the last several decades regarding the optimal extent of surgical resection for low-risk DTC (45, 46). The principal surgical strategies for thyroid resection include thyroid lobectomy with/without isthmusectomy and near-total or total thyroidectomy (8). To date, there has not been a prospective, randomized controlled trial examining optimal extent of surgery for low-risk DTC, and such a trial likely would be prohibitively expensive and difficult to perform, because it has been estimated that between 360 and 800 patients would need to be randomized and then followed for 6 to 10 years (47).

Prior iterations of the ATA guidelines published in 1996, 2006, and 2009 endorsed total or near-total thyroidectomy as the most appropriate initial surgery for DTC >1 cm (9, 22, 48). The evidence cited for these recommendations originated from observational studies demonstrating higher recurrence rates following thyroid lobectomy when compared with total thyroidectomy, total thyroidectomy addressing the risk of multifocal disease in the contralateral thyroid, and total thyroidectomy facilitating more efficient use of RAI in the adjuvant setting and easier surveillance for recurrent disease long term (40, 41, 49–54). Total thyroidectomy, however, obliges lifetime levothyroxine replacement therapy and is associated with higher surgical risk. In an analysis of 62,722 thyroid surgeries from 2003 to 2009 in the Health Care Cost and Utilization Project National Inpatient Survey, Hauch et al. (55) found a higher risk of complications associated with total thyroidectomy (20.4%) compared with thyroid lobectomy (10.8%) (P < 0.0001). Total thyroidectomy was associated with higher risks of hypocalcemia, respiratory complication, hematoma, bleeding, tracheostomy, and recurrent laryngeal nerve injury (55).

In 2007, a pivotal study published by Bilimoria et al. (56) suggested that total thyroidectomy was associated with improved overall 10-year survival compared with thyroid lobectomy. Utilizing the National Cancer Database (NCDB) maintained by the American Cancer Society and the American College of Surgeons, Bilimoria et al. (56) analyzed 52,173 patients with PTC who had surgery between 1985 and 1998. The authors demonstrated a 21% increase in the risk of death for patients with PTC tumors of all sizes (23.9% of the study cohort had PTmC) who underwent thyroid lobectomy compared with total thyroidectomy [hazard ratio (HR), 1.21; 95% CI, 1.02 to 1.44, P = 0.027] after adjusting for demographic, clinical, and pathologic characteristics (56). When survival was compared by tumor size and extent of surgery, survival benefit was not observed for total thyroidectomy patients with PTmCs compared with lobectomy. Total thyroidectomy for patients with PTC ≥1.0 cm was associated with improved 10-year all-cause survival (HR, 1.31; 95% CI, 1.07 to 1.6, P = 0.0009 for lobectomy vs total thyroidectomy) (56). To address potential confounding by larger tumors and more aggressive disease, subgroup analyses were performed for smaller tumors (1.0 to 2.0 cm); a survival advantage again was observed for total thyroidectomy compared with lobectomy (HR, 1.49; 95% CI, 1.02 to 2.17, P = 0.04 for lobectomy vs total thyroidectomy) (56). Recurrence rates were higher when lobectomy was performed for PTC ≥1.0 cm compared with total thyroidectomy (HR, 1.15; 95% CI, 1.02 to 1.30) (56); notably, recurrence is not an outcome reported as part of the participant use file for the NCDB. Despite controlling for tumor and treatment characteristics in the Cox proportional hazards modeling, this study may have been confounded by lack of inclusion of medical comorbidities in the analysis. Of the 8946 patients who underwent thyroid lobectomy alone, 8.7% had PTC tumors >4 cm, 18.4% received RAI despite only having partial thyroid removal, and 1.2% of lobectomy patients had distant metastases (56). This data implies some high-risk PTC pathology was treated with thyroid lobectomy alone and indicates inclusion of patients with other significant clinical factors that likely limited their surgical treatment and may have influenced overall survival (OS).

Haigh et al. (57) used SEER to analyze the outcomes of lobectomy vs total thyroidectomy for both low- and high-risk PTC. Between 1988 and 1995, 5432 patients >20 years who had surgery for PTC were identified; 4402 (81%) had low-risk PTC by age, metastases, extent, and size risk classification criteria (57). In age, metastases, extent, and size low-risk PTC, overall 10-year survival was no different for patients treated with total thyroidectomy compared with lobectomy (89% vs 91%, respectively; P = 0.07) (57). In multivariate analysis, patients with low-risk PTC who underwent total thyroidectomy were actually found to have an increased risk of death when compared with patients who underwent lobectomy (HR, 1.73; 95% CI, 1.28 to 2.33 for total thyroidectomy vs lobectomy) (57). The study, however, was limited by lack of data regarding patients’ comorbidities and by a median follow-up of only 7.4 years, with inclusion of patients who had follow-up for as short as 1 month (57). Subsequent studies using SEER data reinforced these findings, with no observable survival benefit associated with total thyroidectomy compared with thyroid lobectomy. Mendelsohn et al. (58) examined both DSS and OS for 22,724 patients with PTC with a mean follow-up of 9.1 years from 1988 to 2001 in SEER. After adjustment for important clinical and demographic variables, the authors found no difference in 10-year DSS and 10-year OS following lobectomy vs total thyroidectomy (DSS HR, 0.91; 95% CI, 0.71 to 1.15, P = 0.41; OS HR, 0.93; 95% CI, 0.84 to 1.03, P = 0.16 for lobectomy vs total thyroidectomy) (58). Examining a SEER cohort from 1983 to 2002, Barney et al. (59) again found no significant 10-year OS (HR, 1.07; 95% CI, 0.91 to 1.25, P = 0.431 for lobectomy vs total thyroidectomy) or DSS advantage (HR, 0.77; 95% CI, 0.53 to 1.12, P = 0.176) among patients treated with total thyroidectomy compared with thyroid lobectomy after adjusting for other factors that could be associated with high-risk PTC.

In 2014, Adam et al. (60) analyzed survival for 61,775 patients from 1998 to 2006 in the NCDB with PTC tumors 1.0 to 4.0 cm. With a median follow-up of 82 months, the authors demonstrated no significant OS advantage for patients undergoing total thyroidectomy compared with lobectomy (HR, 0.96; 95% CI, 0.84 to 1.09 for total thyroidectomy vs lobectomy) (60). The NCDB database began more accurately collecting data about patient comorbidities starting in 2003, as well as data regarding ETE and multifocality in 2004. More medical comorbidity as indicated by a Charlson Comorbidity Index ≥2 was associated with OS in this study cohort (HR, 3.83; 95% CI, 3.05 to 4.81 vs Charlson Comorbidity Index = 0) (60), so the benefit of total thyroidectomy for OS described by Bilimoria et al. (56) in 2007 was likely confounded by inability to adjust for this important covariate in their modeling.

Patient age <45 years has been a significant inflection point in prognostic staging for patients with DTC; according to the AJCC, patients aged <45 years are considered to have a lower risk of cancer-specific death regardless of tumor size and cervical lymph node status unless they also have distant metastatic disease, in which case they are upstaged to stage II (3). However, guidelines regarding extent of surgery have not differed based on patient age (9, 22, 48). Utilizing both the NCDB (1998 to 2006) and SEER database (1988 to 2006), Adam et al. (61) studied 43,032 adult patients aged <45 years who had PTC tumors 1 to 4 cm (29,522 NCDB, 13,510 SEER) with a median follow-up of 83 months (NCDB) and 115 months (SEER). In multivariate logistic regression analysis, OS was similar for young patients treated with lobectomy and total thyroidectomy (NCDB HR, 1.45; 95% CI, 0.84 to 2.51 for total thyroidectomy; SEER HR, 0.95; 95% CI, 0.70 to 1.29 for total thyroidectomy). These data support the notion that thyroid lobectomy and total thyroidectomy are associated with comparable long-term outcomes for PTC ≤4 cm, even among young patients.

Unfortunately, recurrence as an outcome is not collected currently in large national administrative and claims databases in the United States, leaving research regarding the extent to which thyroid lobectomy adequately controls the incidence of tumor recurrence or need for reoperation when compared with total thyroidectomy limited to single institutional experiences. Nixon et al. (62) reviewed the Memorial Sloan-Kettering clinical experience with DTC from 1986 to 2005 based on 10-year thyroid bed recurrence, neck recurrence, distant recurrence, OS, and DSS; median follow-up was 99 months. There were 889 patients with T1 (1 to 2 cm) or T2 (2 to 4 cm) intrathyroidal tumors, of which 361 underwent lobectomy (40.6%) and 528 total thyroidectomy (59.4%); most (90%) had PTC on histology (62). Of note, 193 of the lobectomy patients had either intermediate- or high-risk disease by Memorial Sloan Kettering Cancer Center (MSKCC) criteria (intermediate defined by a patient <45 years with a high-grade tumor or a patient >45 years with a low-grade tumor, and high risk defined by a patient >45 years with a high-grade tumor) (63). There was no difference in 10-year thyroid bed or neck recurrence rates when comparing patients treated with lobectomy vs total thyroidectomy (0% lobectomy vs 0% total thyroidectomy for thyroid bed recurrence, 0% lobectomy vs 0.8% total thyroidectomy for neck recurrence) (62).

A large study from the Ito Hospital in Japan examined 1088 patients with PTC from 1986 to 1995 who underwent thyroid lobectomy as definitive treatment (64). Patients did not receive RAI ablation and routinely had both a central and lateral compartment neck lymph node dissection performed as part of the operative strategy, even in the absence of clinically apparent lymphadenopathy. The study results were limited, in that there was no comparison total thyroidectomy treatment group, and some study patients had intermediate- and high-risk features [7.7% had ETE and 10% had tumors >4 cm] (64). Despite these limitations, the study authors found that the rate of recurrence within the thyroid remnant at 25 years was 6.5%, and the rate of recurrence in regional lymph nodes was 9.4%; both rates are comparable to those found in the United States following total thyroidectomy (65). Factors associated with regional lymph node recurrence were high-risk tumors with ETE, positive lymph nodes, and tumors >4 cm (64). Taken as a whole, thyroid lobectomy for low-risk PTC patients afforded excellent long-term outcome (100% 25-year DSS) and with a low risk of thyroid remnant (6.0%) or regional lymph node recurrence (6.3%) (64). In patients without previous history of cervico-facial radiation or a strong family history of thyroid cancer, the ATA guidelines currently endorse thyroid lobectomy as likely sufficient extent of surgery for DTC tumors <4 cm with no intermediate- or high-risk features (8). In many cases where DTC is diagnosed on final pathology, removal of the thyroid remnant (completion thyroidectomy) is not needed unless total thyroidectomy would have been recommended preoperatively when informed by the final histology (8).

Thyroid lobectomy appears to be cost-effective compared with total thyroidectomy for low-risk PTC. Utilizing a Markov decision tree model and a reference case of a 40-year-old female with a low-risk, 2.5-cm PTC, Lang et al. (66) found that the cost of thyroid lobectomy ($30,303.22) in US dollars based on 2016 Medicare reimbursement discounted 3% annually over a 25-year period was higher than that of total thyroidectomy ($29,530.93). The model used recurrence data from the 2007 Bilimoria study (56) and assumed an annualized loco-regional recurrence rate of 0.98% for lobectomy and 0.77% for total thyroidectomy (66). Of the patients undergoing thyroid lobectomy, 42.4% were assumed to have high-risk features on final pathology that would require completion thyroidectomy in the model ($10,364) (66). Thyroid lobectomy afforded more QALYs than total thyroidectomy over the same 25-year time frame (16.046 QALYs vs 15.747 QALYs, respectively), and the ICER of $2,577.65 was below a recommended threshold of $50,000 per QALY (66).

Molecular markers and extent of surgery

The role of molecular testing in the management of DTC has been primarily in the evaluation of indeterminate thyroid cytology where a diagnosis of thyroid cancer is not immediately evident (67, 68). Because most low-risk DTC diagnosed based on cytology alone or with adjunctive molecular testing now can be managed with thyroid lobectomy or total thyroidectomy, it is becoming less clear to what extent molecular testing will serve as a guide with regard to the optimal extent of surgery. One significant point of controversy is centered around the impact of the BRAF V600E mutation on risk stratification and informing aggressiveness of surgery or adjunctive treatment. In several studies, BRAF V600E has been associated with an increased risk of recurrence (69–71), as well as clinical and pathologic features indicative of high-risk DTC, including lymph node metastases, ETE, and distant metastasis (69). However, it has not been shown to independently be associated with disease-specific mortality (72). In a multinational, multi-institutional retrospective study of 1849 patients with a median follow-up of 33 months, Xing et al. (72) found BRAF V600E mutation–positive thyroid cancer was associated with decreased mortality overall when compared with BRAF wild type [12.87 deaths per 1000 person years BRAF V600E(+) vs 2.52 deaths per 1000 person years BRAF wild type]. However, with adjustment for patient age, sex, ETE, lymph node metastasis, and distant metastasis, BRAF V600E was not found to be independently associated with mortality (HR, 1.21; 95% CI, 0.53 to 2.76) (72). In a different study examining the role of BRAF V600E in the context of low-risk DTC (defined as T1a-T2N0M0), Elisei et al. (71) found mutated BRAF to be independently associated with an increased rate of recurrence at 5 years [16% BRAF V600E(+) vs 3.3% BRAF V600E(–), P < 0.0001]. Although this study was intended to focus on low-risk cancers, it did include some aggressive PTC variants (13.2% of cases) and cases with evidence of vascular invasion (7.2%), which would be assigned to an ATA intermediate-risk category (8, 71).

“Thyroid lobectomy appears to be cost-effective compared with total thyroidectomy for low-risk PTC”

The cost utility of molecular testing for determining extent of thyroid surgery has been examined. Utilizing a decision tree model, Yip et al. (73) found that the addition of molecular testing to thyroid cytology results increased the rate of initial total thyroidectomy (18.2% with molecular testing, 16.1% without molecular testing) and provided an overall cost savings by reducing the incidence of a two-stage thyroid operation (lobectomy followed by completion thyroidectomy). The model assumed that any mutation-positive result should be treated with surgery given the high likelihood of malignancy, and total thyroidectomy was performed for all thyroid cancer cases in accordance with 2009 ATA guidelines (73). However, many cases of thyroid cytology with mutation-positive results were the follicular variant of PTC and represented low-risk DTC (74, 75).

Role of prophylactic central neck dissection

The central neck (level VI) is the pretracheal and paratracheal area immediately surrounding the thyroid bordered by the hyoid bone superiorly, the carotid arteries laterally, and the axial plane of the innominate artery inferiorly (76). The lymph nodes within this compartment are often involved with lymph node metastasis from DTC, and comprehensive surgical dissection should be done in the setting of clinically apparent lymph node metastasis (8). The practice of prophylactic central compartment lymph node dissection (pCLND) in the context of clinically node negative DTC is contentious (77), especially in the context of low-risk disease. Indeed, the most recent ATA guidelines recommend consideration of pCLND for patients who are at an intermediate or high risk of recurrence (tumors >4 cm, tumors with ETE, and/or lateral neck lymph node disease) or if the information gained from pCLND would inform any adjuvant treatment strategy (8). Routine use of pCLND is in part controversial because of the potential added risk associated with the procedure, including transient or permanent hypoparathyroidism and recurrent laryngeal nerve injury. These increased risks need to be weighed against the potential benefits, which might include reducing the risk of death and recurrence and advising future treatment and surveillance (i.e., informing the use of RAI or reducing thyroglobulin levels) (77–83). The possibility of answering whether pCLND is necessary with a prospective randomized controlled study is not likely, as it has been estimated that 5480 patients with 5 years of follow-up would be needed to achieve statistical power (84).

The recent evidence whether pCLND impacts DSS is mixed. In a retrospective study of 640 patients (282 patients without pCLND, 358 with pCLND) with mean follow-up of 10.5 years, Barczyński et al. (85) demonstrated that 10-year DSS was improved with pCLND, from 92.5% for patients without pCLND to 98% for patients with pCLND (P = 0.034). However, the pCLND study group was a more contemporary cohort (1998 to 2002 vs 1993 to 1997 without pCLND) that included a larger number of pCLND patients who were upstaged by a finding of nodal metastasis; as a result, RAI was used more often in the pCLND group (64.5% with vs 28.0% without pCLND, P < 0.001) (85). On multivariate analysis, the only factors associated with compromised survival were ETE and distant metastases (85). Therefore, benefit of pCLND for patients with low-risk disease was unclear. In a shorter, retrospective single-institution study of 1798 patients without pCLND treated at the Memorial Sloan-Kettering Cancer Center from 1984 to 2010, Nixon et al. (86) demonstrated that 5-year DSS was 100% with a median follow-up of 46 months. These findings were notwithstanding inclusion of 732 patients (40.7%) who had ATA intermediate- or high-risk cancers, again suggesting that there is likely no measurable benefit from pCLND in the setting of low-risk disease (86).

The potential benefit of pCLND for reducing the risk of loco-regional recurrence for low-risk DTC is difficult to discern. In a large meta-analysis representing 14 studies and 3331 patients with PTC, Lang et al. (87) reported a 35% reduction in the risk of loco-regional recurrence with pCLND (pooled incidence rate ratio, 0.65; CI, 0.48 to 0.86). However, ETE was reported in 12.8% to 58.5% of the patients treated without pCLND in 9 of 14 studies, indicating that a significant proportion of the study subjects did not have low-risk disease (87). In another meta-analysis that included 11 studies and 2318 patients, Wang et al. (88) showed a trend toward fewer recurrences with pCLND (4.7% vs 7.9% without pCLND) that did not reach statistical significance [relative risk (RR), 0.59; 95% CI, 0.33 to 1.07]. It was estimated that as many as 31 patients would need to be treated with pCLND to reduce the risk of recurrence by one case. In summary, reduction in the risk of recurrence afforded by pCLND for low-risk DTC is not immediately discernable.

The cost-effectiveness of routine pCLND is similarly not clear. In one study by Wong et al. (89), Markov decision tree analysis was used with an index case of a 50-year-old female with low-risk, 1.5-cm PTC and found an extra cost of only $34.52 for routine pCLND with an addition of 0.32 QALY per patient. With an assumption that the rate of recurrence requiring reoperation with pCLND is 0.87% annually and 1.57% without pCLND, ICER fell below the willingness-to-treat threshold of $50,000 at 9 years of follow-up and was cost-effective (89). A second study by Garcia et al. (90) used a 40-year-old female reference patient with a low-risk, 2.0-cm PTC and assumed 1.8% probability of lifetime recurrence requiring reoperation for pCLND patients and 2.3% for patients without pCLND. Importantly, all patients treated without pCLND were given additional treatment with RAI, and only patients with lymph node disease treated with pCLND were treated with adjuvant RAI. Total thyroidectomy without pCLND cost $5,763 less and was associated with greater effectiveness (0.03 QALY per patient), and pCLND was found to not be cost-effective (90).

Choosing the surgeon

On average, superior patient outcomes following thyroid surgery are observed when the procedures are performed by high-volume thyroid surgeons, independent of specialty (91, 92). The ATA guidelines currently comment on this topic, acknowledging improved outcomes with referral of patients to high-volume surgeons; however, no specific recommendation is given nor is a clear threshold specified in what would be considered a high-volume thyroid surgeon (8). In a recently published study by Adam et al. (93), data from the Health Care Utilization Project-National inpatient Sample for 16,954 patients undergoing total thyroidectomy by 4627 surgeons between 1998 and 2009 found that more than one-half (51%) of surgeons in the United States who had done thyroid surgery performed only one total thyroidectomy per annum during the study period. Utilizing a multivariate logistic regression model with restricted cubic splines, the authors determined that increasing surgeon volume was associated with a reduced probability of experiencing a complication in a dose-dependent manner up to 26 thyroid surgery cases per year (93). Overall, 6% of patients experienced at least one complication from total thyroidectomy, with the most common complications related to hypoparathyroidism and nerve injury (2% overall) (93). Total thyroidectomy by high-volume surgeons defined by >25 surgeries per annum was associated with fewer complications (4.1% high volume vs 6.4% low volume, P < 0.0001), lower median cost ($5826 high volume vs $6385 low volume, P < 0.0001), and shorter median length of hospital stay (1 day high volume vs 2 days low volume, P < 0.0001) (93). After adjusting for patient demographics, comorbidities, diagnosis, and hospital characteristics, patients treated by low-volume surgeons were 51% more likely to incur a complication [Odds ratio (OR), 1.51; CI, 1.16 to 1.97; P = 0.002] and longer hospital stay (12% increase; CI, 3% to 21%; P = 0.006) (93). This should serve as a first step for establishing a minimum threshold requirement for choosing a surgeon and should be replicated in another data set. It is concerning that the majority of patients undergoing total thyroidectomy in the United States today are having their surgery performed by low-volume surgeons and therefore are potentially exposed to excessive risk. As 81% of patients had surgery by a low-volume surgeon, the societal cost implications are substantial, and the findings have important implications for patient referrals, reimbursement, and graduate surgical education in the current era of value-based care.

Balancing benefits-risks and addressing limitations of available data

As outlined, the risks of thyroid surgery are not trivial, and the possibility of complication becomes greater with increasing extent of surgery, from lobectomy to lobectomy with pCLND to total thyroidectomy to total thyroidectomy with pCLND. In the context of this review with a focus on low-risk DTC, the risk of recurrence is low, and in the case of PTmC, the risk of tumor progression without any treatment is relatively low. These surgical risks and the disadvantage for 22% of lobectomy patients needing thyroid hormone treatment of hyposufficiency and 100% of total thyroidectomy patients needing full replacement are balanced against data that show little benefit for recurrence and survival for total thyroidectomy when compared with lobectomy (31). The limitation of these data are related to a lack of large, prospective, randomized studies with long-term follow-up. Much of the single-institution retrospective data are limited by short follow-up and inadequate number of patients to afford the statistical power needed to control for multiple important covariables. The large national administrative and claims databases lack granularity and accurate longitudinal cancer recurrence information. Although increased surgical volume mutes some of the increase in complications, even in the hands of high-volume thyroid surgeons, total thyroidectomy is associated with twice the risk of any complication (55). Similarly, the benefit of pCLND in reducing recurrence or prolonging survival is not clear, and the data available are largely biased by the most significant studies reflecting outcomes from high-volume surgeons at high-volume thyroid surgery centers. The reality is that the majority of thyroid surgery is done by low-volume surgeons, and access for many patients to high-volume surgeons is limited. With the general de-escalation of invasive treatment of low-risk DTC over the last decade, the balance of surgery risk weighs heavier than the benefit gained and has been reflected in the evolution of guideline recommendations (Table 4). The least surgical risk is associated with active surveillance; however, there may be real limitations and hurdles in implementing active surveillance programs for PTmC outside of Japan (39).

It is unlikely that definitive large, prospective, multi-institutional, randomized controlled surgical trials can be executed to resolve the current controversies in the surgical management of low-risk DTC due to their anticipated large size and cost. However, current national cancer databases could be enhanced to provide information about preoperative cytology, important thyroid surgery–specific outcomes, such as recurrent laryngeal nerve (RLN) dysfunction, hypoparathyroidism, and more accurate coding pertaining to the intent and extent of surgery. Surgical pathology could be reported in a manner that is better aligned with the ATA risk stratification system for recurrence, and molecular testing results could be included. Ideally, data regarding response to therapy would include more meaningful outcome data than mortality, like recurrence. As the breadth of molecular testing expands, it will be vital to conduct rigorous prospective, multi-institutional studies that examine the utility of the tests in combination with thyroid cytology and ultrasound to more accurately distinguish low-, intermediate-, and high-risk thyroid cancers to better inform surgical decision-making with regard to appropriate extent of surgery.

“As 81% of patients had surgery by a low-volume surgeon, the societal cost implications are substantial”

In an era of patient-centered care and comparative effectiveness, more robust multi-institutional data are needed about patient-reported outcomes. In particular, it will be important to expand our understanding of how patients make decisions based on their specific values and preferences, especially in scenarios where there is evidential equipoise, such as the appropriate extent of surgery for low-risk DTC.

RAI

RAI treatment vs no RAI treatment

Historically, near-total or total thyroidectomy followed by RAI has been standard of care for patients with DTC. RAI is taken up by the sodium-iodine symporter and concentrated in the thyroid follicular cells. RAI can be used to treat suspected or known iodine-avid residual or metastatic disease in intermediate- or high-risk DTC patients. It also can be used to eliminate the postsurgical thyroid remnant in patients with low-risk DTC.

In 2011, Haymart et al. (23) evaluated use of RAI in 189,219 patients with DTC treated at 981 hospitals associated with the NCDB between 1990 and 2008. Over time, there was an increase in the proportion of patients receiving RAI across all tumor sizes (P < 0.001). In female patients <45 years of age with tumors <1.0 cm and stage I disease, there was tremendous hospital-level variation in the utilization of RAI. Based on data from a contemporary cohort of 85,948 patients treated from 2004 to 2008, they found that 21.1% of the variation in use of RAI was accounted for by patient and tumor characteristics and 17.1% by hospital type and case volume. Unexplained hospital characteristics accounted for 29.1% of the variance (23). In a follow-up survey study of endocrinologists and nuclear medicine physicians treating thyroid cancer, Papaleontiou et al. (30) identified factors influencing whether a patient receives RAI. Extent of disease, adequacy of surgical resection, patient willingness to receive RAI, and patient age (which has implications for staging) were the most important factors in RAI decision-making. Less expected, physician and patient worry about death were also important in physician decision-making, with low-volume physicians significantly more likely to report patient (P < 0.001) and physician (P = 0.016) worry about death as important. Accepted standard at the affiliated hospital (P = 0.020) and beliefs expressed by colleagues (P = 0.003) were also more important to low-volume physicians (30). These studies on variation in practice patterns highlight the lack of consensus over use of RAI for low-risk DTC. This lack of physician consensus has implications for patients, who report receiving conflicting messages on the appropriateness of RAI (94).

Several studies have attempted to better define the role of RAI in thyroid cancer management, including assessing whether use of RAI in low- or intermediate-risk disease impacts survival. Ruel et al. (95) found that for 21,870 patients with intermediate-risk PTC, defined as stage T3, N0, M0/Mx and T1-3, N1, M0/Mx, after multivariate adjustment for demographic and clinical factors, RAI was associated with a 29% reduction in risk of death (HR, 0.71; 95% CI, 0.62 to 0.82). However, the benefit of RAI for low-risk DTC is less clear. The National Thyroid Cancer Treatment Cooperative Study (NTCTCS) Group studied the effect of administration of postoperative RAI for DTC as adjuvant therapy and for treatment of disease at 11 institutions. In 2006, the study cohort included 2936 patients with DTC followed for a median of 3 years, and in 2015, it included an updated cohort of 4941 patients with DTC followed for a median of 6 years (96, 97). The NTCTCS has its own staging system, which is similar but not identical to the AJCC TNM staging system. In the 2006 cohort, 70% of the patients were considered low risk by NTCTCS staging compared with 73% by AJCC TNM staging. In patients with NTCTCS stage II disease, postoperative administration of RAI was associated with significantly improved OS (RR, 1.71; 95% CI, 1.07 to 2.74), but there was no improvement in DSS (RR, 1.21; 95% CI, 0.26 to 3.92) or disease-free survival (DFS; RR, 1.03; 95% CI, 0.75 to 1.39) (96). In contrast, in stage I patients, postoperative RAI was associated with compromised OS (RR, 0.0006; lower limit of 95% CI approaches 0). Of the seven patients who died, only one died secondary to thyroid cancer. DFS appeared to be worse in patients with stage I disease who received RAI (RR, 0.64; 95% CI, 0.47 to 0.85). However, with propensity score analysis, there was no significant difference in DFS in stage I patients receiving RAI vs not receiving RAI (96). In the 2015 study, administration of RAI in patients with NTCTCS stage I disease was again associated with a decrease in DFS, but unlike the 2006 cohort, there was no difference in OS. Propensity score analysis again showed no difference. In addition, with the updated 2015 data, the improved OS observed among stage II patients was no longer significant (97). In summary, the NTCTCS data suggest no clear benefit of RAI in the treatment of low-risk DTC.

A study by Schvartz et al. (98) evaluated use of RAI among 1298 DTC patients treated between 1975 and 2005. For this study, “low risk” was defined as tumors <4 cm and with no regional or distant metastases. With a median follow-up of 10.3 years, only 19 patients experienced recurrences, with 15 (1.6%) occurring in patients who received RAI and 4 (1%) in patients who did not receive RAI. After propensity score stratification, there was no significant difference in OS or DFS among patients who did vs did not receive RAI (98).

A recent study by Kim et al. (99) evaluated 704 patients with PTmC diagnosed between 1994 and 2004. In this study, if the patients with PTmC had microscopic ETE, lymph node metastases, or multifocality, they were considered intermediate risk. A total of 578 (82%) patients received RAI, and 126 (18%) did not receive RAI. The study cohort was considered disease free after initial therapy. With a median follow-up of 5.3 years, there were six recurrences (0.9%). All six patients had received RAI. With propensity score analysis, there was no difference in likelihood of recurrence between the patients who did and did not receive RAI (P = 0.17) (99).

In addition to risk stratifying based on tumor characteristics, additional studies have looked at response to therapy reclassification, and specifically the role of postoperative thyroglobulin in determining which patients are likely to benefit from RAI (6, 100–102). In a single institution study by Ibrahimpasic et al. (102), a total of 424 patients were status post total thyroidectomy for PTC and had a postoperative thyroglobulin <1 ng/mL. Low risk was defined by GAMES criteria (grade, age, distant metastasis, ETE, and size of neoplasm), and a total of 80 (19%) patients were defined as low risk. In the low-risk cohort, 35 (44%) patients received postoperative RAI, and 45 (56%) did not. There were no disease-specific deaths in the low-risk cohort, and there was one recurrence that occurred in the group that did not receive RAI. Recurrence-free survival was similar in those patients who did and did not receive RAI (96% vs 100%, respectively; P = 0.337), suggesting treatment with RAI may be unnecessary.

Rosario et al. (103) evaluated 136 low-risk PTC patients status post total thyroidectomy without RAI treatment. For this study, low risk was defined as T1b-T3 N0 M0. In this study, all patients had a stimulated thyroglobulin ≤1 ng/mL in the setting of negative antithyroglobulin antibodies and a normal neck ultrasound approximately 4 months after surgery. With a median follow-up of 44 months, 134 (98.5%) continued to have serum thyroglobulin ≤1 ng/mL, negative antithyroglobulin antibodies, and normal neck ultrasound. One patient developed lymph node metastases, and another had an increase in thyroglobulin antibodies without evidence of recurrence. The authors concluded that patients with low-risk PTC with a stimulated thyroglobulin ≤1 ng/mL do not require remnant ablation (103).

In a subsequent study, Rosario et al. (104) studied 154 patients with low-risk PTC, negative neck ultrasound, negative thyroglobulin antibodies, and unstimulated thyroglobulin ≤0.25 ng/mL after total thyroidectomy. These patients subsequently had a whole-body scan (WBS) and iodine-131 (I-131) ablation. None of the patients had ectopic uptake on WBS, and 9 to 12 months after ablation, 150 (97.4%) had a stimulated thyroglobulin ≤1 ng/mL. With a median follow-up of 2 years, there were no PTC recurrences. The authors concluded that RAI remnant ablation may not be needed in low-risk patients with PTC >1 cm, nonstimulated postoperative thyroglobulin ≤0.25 ng/mL, negative thyroglobulin antibodies, and negative neck ultrasound (104).

Orlov et al. (105) studied 129 patients with PTC status post total thyroidectomy and selective central neck dissection. All patients were defined as having low or intermediate disease based on PTC confined to the thyroid and central neck lymph nodes. In this study, postoperative stimulated thyroglobulin was used to risk stratify. Of the 129 patients, 84 (65%) had a stimulated thyroglobulin <1 ng/mL, 40 (31%) between 1 and 5 ng/mL, and 5 (4%) >5 ng/mL. Patients with a stimulated thyroglobulin <1 ng/mL were discouraged from receiving RAI, whereas patients with a stimulated thyroglobulin >5 ng/mL were encouraged to proceed with RAI. The 40 patients with thyroglobulin between 1 and 5 ng/mL and with a negative neck ultrasound were followed but informed that they may need to receive RAI in the future. Of these patients, 8 (20%) ultimately received RAI secondary to higher thyroglobulin or rising thyroglobulin levels. All patients were followed prospectively for a mean of 6.2 years. Only one of the 13 patients who received RAI experienced a recurrence. The authors advocated the benefit of using postoperative stimulated thyroglobulin to determine when to treat with RAI (105).

Although a stimulated or unstimulated thyroglobulin level that is <1 ng/mL does not rule out residual disease that would benefit from RAI treatment, it is agreed that a thyroglobulin level of 5 to 10 ng/mL makes the likelihood of residual disease more probable (102, 103, 105–108). The data suggest that in patients with low- or intermediate-risk disease, there is no difference in the likelihood of recurrence with or without RAI if a patient has an unstimulated thyroglobulin that is <1 to 2 ng/mL (102, 103, 108). If the stimulated thyroglobulin is above 5 ng/mL, RAI may be beneficial; if the level is 1 to 5 ng/mL, then additional clinical factors should be used to help determine whether it is needed; and if the level is <1 ng/mL, then RAI remnant ablation is unnecessary (105).

“It is agreed that a thyroglobulin level of 5 to 10 ng/ ml makes the likelihood of residual disease more probable”

RAI administration

If RAI is deemed to be necessary, there is also a lack of consensus regarding mechanisms surrounding RAI preparation and associated imaging. In a 2013 survey study by Haymart et al. (24) evaluating treatment patterns, 234 (48%) endocrinologists and nuclear medicine physicians used a pretreatment scan prior to selecting RAI dose vs 250 (52%) who treat without a pretreatment scan. Some also advocate the use of single-photon emission computed tomography/computed tomography (SPECT-CT) with the pretreatment scan. The benefits of preablation radioiodine scans with SPECT/CT include determining whether atypical uptake is thyroid cancer vs physiologic or nonthyroid in origin and identification of regional and distant metastases prior to radioiodine therapy, which can lead to a change in management (109, 110). However, this additional imaging is associated with more costs and may only benefit select patients, such as those with higher postoperative thyroglobulin levels (111). If a pretreatment scan is used, there is also debate about whether low-dose I-131, low-dose iodine-123 (an alternative radioisotope associated with less stunning), or technetium-99m is preferred (112–114). Although controversy exists about the role of pretreatment scans, it is generally agreed that if RAI is administered, then a post-therapy scan (with/without SPECT-CT) is useful for determining the RAI avidity of residual structural disease (8).

In addition to debates about the imaging associated with RAI administration, preparation with recombinant human thyroid-stimulating hormone (rhTSH) vs thyroid hormone withdrawal remains a controversial topic. Observational data and six randomized clinical trials suggest that rhTSH is as effective as withdrawal for remnant ablation in low-risk DTC (115–122). Two of the recent phase 3 clinical trials assessed both thyroid hormone withdrawal vs rhTSH and low-dose (30 mCi) vs high-dose (100 mCi) RAI for remnant ablation (117, 118). Schlumberger et al. (117) enrolled 752 patients with low-risk DTC between 2007 and 2010. Of the DTC patients, 92% had PTC. Low risk was defined as PTmC with lymph node involvement or unknown lymph node status, PTC >1 to 2 cm and any lymph node involvement, and PTC >2 cm but ≤4 cm without lymph node involvement (117). Adequate ablation was defined as a normal neck ultrasound and level of rhTSH-stimulated thyroglobulin ≤1 ng/mL, or if thyroglobulin antibodies were detectable, then a normal I-131 total-body scan. In this trial, ablation was adequate in 91.7% receiving rhTSH vs 92.9% of those receiving thyroid hormone withdrawal. A parallel study by Mallick et al. (118) enrolled 438 patients. This study had similar criteria, as low-risk DTC was defined as tumor <4 cm with the possibility of lymph node involvement but no distant metastases (118). Adequate ablation was defined as both a negative I-131 scan and thyroglobulin level <2.0 ng/mL. If both criteria were not available, then just one was used. If antibodies were present, immunometric and radioimmunoassay concordance was compared. Ultimately, 87.1% of the patients treated with rhTSH vs 86.7% of the cohort undergoing withdrawal from thyroid hormone experienced adequate ablation (118). In both these trials, ablation rates were comparable regardless of whether withdrawal or recombinant therapy strategies were taken.

In addition, both trials found that symptoms of hypothyroidism were significantly more common with thyroid hormone withdrawal, and this was associated with compromised quality of life (117, 118). Additional studies have found that short-term hypothyroidism after thyroid hormone withdrawal is associated with compromised quality of life compared with rhTSH (122–124). An additional benefit of rhTSH is reduced radiation exposure to the body as compared with withdrawal (125). In a prospective, randomized, controlled, multinational trial, Hänscheid et al. (125) evaluated 63 patients status post thyroidectomy randomized to 100 mCi of I-131 with withdrawal vs with rhTSH and found that the effective half-time of I-131 in the remnant thyroid tissue was longer with rhTSH (P = 0.01), but the dose to the blood was significantly lower (P < 0.0001). However, despite its strengths, rhTSH is more costly and, in some scenarios, may not be cost-effective (124, 126). Therefore, although similar efficacy for low-risk disease, controversy over the preferred preparation still exists.

Optimal dose of RAI for remnant ablation

Once a decision is made to administer RAI for remnant ablation, then optimal dose should be considered. As shown in Table 3, prior to the 2015 ATA guidelines, either 30 mCi or 100 mCi of I-131 was recommended for remnant ablation (8, 9, 22). However, since the 2009 ATA guidelines were published, several randomized controlled trials assessed dose of RAI for remnant ablation, and the majority found similar efficacy between high- and low-dose RAI (117, 118, 127–129). The two recent randomized phase 3 clinical trials that evaluated withdrawal vs rhTSH also addressed the topic of RAI dose (117, 118). Schlumberger et al. (117) evaluated 30 mCi vs 100 mCi of I-131 among 752 patients with low-risk DTC. Ablation was considered adequate in 92% of the patients, with equivalence shown between low- vs high-dose RAI. The study by Mallick et al. (118) evaluated the same doses of RAI (30 mCi vs 100 mCi) among 438 patients; adequate ablation was demonstrated in 85% of the group receiving low-dose RAI vs 88.9% in the group receiving high-dose RAI. Noninferiority of low-dose RAI was demonstrated; in fact, the study by Mallick et al. (118) found that higher dose was associated with significantly more adverse effects (e.g., nausea and neck pain). In 2015, the ATA guidelines recommended that if RAI is used for remnant ablation, 30 mCi is favored over higher doses (8).

Balancing benefits-risks and addressing limitations of available data

RAI for remnant ablation facilitates long-term follow-up and adds a sense of security for many physicians and patients. In a survey of 486 physicians from Canada and the United States, Sawka et al. (130) found that strong physician recommendations for RAI were associated with physician beliefs that treatment with RAI decreases mortality and recurrence and facilitates follow-up. Physicians without strong convictions about RAI were more likely to incorporate patient preference (130). Sawka et al. (131) later administered a Decision Regret Scale questionnaire to 44 patients, with 26 of the 44 having received RAI. Although there was no significant difference in regret between patients who did or did not receive RAI, patients involved in the treatment decision had less regret than those not involved. Sawka et al. (132) subsequently developed a computerized decision aid on adjuvant RAI for patients with low-risk DTC. In 74 low-risk patients, use of adjuvant RAI did not differ between the cohort who received the decision aid vs the controls (P = 0.278). However, medical knowledge was improved (P < 0.001) and decisional conflict reduced (P < 0.001) (132). This work illustrates the importance of patient education and involvement when determining whether to use RAI for remnant ablation in low-risk DTC.

The decision to treat with RAI is closely linked to decisions on extent of surgery. In addition to RAI not being recommended in patients who receive lobectomy, more extensive surgery is associated with greater use of RAI (29, 81). In a survey of 560 surgeons, Haymart et al. (29) found that surgeons who recommend total thyroidectomy over lobectomy for a subcentimeter unifocal PTC were significantly more likely to recommend pCLND for PTC regardless of tumor size (P < 0.001) and also more likely to favor RAI in patients with unifocal PTC ≤1 cm (P = 0.001), PTC 1.1 to 2 cm (P = 0.004), and intrathyroidal multifocal PTC ≤1 cm (P = 0.004). This data suggest that the relationship between extent of surgery and use of RAI is related to some physicians having a general preference for more intensive treatment of low-risk DTC across all phases of disease management. In comparison, Wang et al. (81) evaluated the records of 103 patients who underwent total thyroidectomy for DTC between 2009 and 2010; 49 (48%) had pCLND, and of these patients, 20 (41%) had positive cervical lymph nodes. pCLND led to a change in recommendations for RAI in 14 (70%) patients. Thus, the study by Wang et al. suggests that more extensive surgery uncovers more disease and therefore leads to more treatment with RAI. Because the relationship between surgery and adjuvant therapy is complex and multifactorial, Carty et al. (133) from the ATA Surgical Affairs Committee have emphasized the importance of interdisciplinary communication to improve patient care and reduce inappropriate treatment.

RAI is not without risks or costs. With high doses of RAI, patients are at risk for sialadenitis, which can lead to dry mouth, increased caries, painful swelling of the cheek, and lacrimal duct damage, which is associated with excess tearing (134–137). High doses of RAI also have been shown to be associated with increased risk of development of a second primary malignancy (138–140). Low-risk DTC patients should receive lower doses of RAI, and therefore the associated risks should be less. However, Iyer et al. (140) found that even in patients with low-risk DTC (T1, N0), RAI was associated with risk of second primary malignancy [standardized incidence ratio (SIR), 1.21; 95% CI, 0.93 to 1.54]. Risk of salivary gland malignancies (SIR, 11.13; 95% CI, 1.35 to 40.2) and leukemia (SIR, 5.68; 95% CI, 2.09 to 12.37) were significantly elevated with risk of leukemia significantly greater in patients aged <45 years (SIR, 5.32; 95% CI, 2.75 to 9.30) (140).

Goffredo et al. (141) used the NCDB database to study use of RAI for PTmC. Despite guideline recommendations against RAI, they found that 14,146 (23.3%) patients with PTmC were inappropriately treated with RAI. Patients who were younger, white, managed at nonacademic centers, and had multifocal or larger tumors were more likely to receive RAI (P < 0.001). It was estimated that this inappropriate use of RAI for PTmC was associated with a cost burden of $9,499,497 to $16,098,824 per year at a national level (141).

There are limitations to the studies evaluating whether remnant ablation is necessary for low-risk DTC. One limitation is that the definition of low-risk varies by study. This makes comparisons between studies difficult. Second, most studies are retrospective cohort studies, and for many of these studies, whether RAI is used for remnant ablation or treatment of residual disease is unclear. Third, many studies focus on OS, and OS may not be the optimal end point for a cancer with an excellent prognosis, as most patients with DTC ultimately die of diseases other than their thyroid cancer. Recurrence is not captured by large cancer registries (142), and even when data on DSS and recurrence are available, the event rate is low, thus limiting the quality of many single-institution studies. In addition, these single-institution studies are susceptible to ascertainment bias, as the patient cohort, treatment pattern, and follow-up may differ from the population at large.

The number of patients needed for enrollment and the length of follow-up necessary prohibit development of a clinical trial to evaluate the utility of RAI for low-risk DTC. However, the low event rate and slow progression suggest that in low-risk DTC, the potential benefits of RAI are modest at most. In addition, the excellent prognosis suggests the safety of a “watch and wait” approach prior to administering RAI. Thus, there is both the need and ability to perform prospective studies evaluating outcomes with no RAI treatment. Because there tends to be strong differences in both physician preference and assessment of risks-benefits, there is a need for educational tools for patients and physicians to reduce unwarranted variation in care.

“The low event rate and slow progression suggest that in low-risk DTC, the potential benefits of RAI are modest at most”

In the 2015 ATA guidelines, there is a strong recommendation not to treat with RAI if the patient has unifocal PTmC. Secondary to limitations of available data, the ATA guidelines give a weak recommendation to avoid RAI in the setting of multifocal PTmC and other low-risk DTC. For most low-risk DTC, patient preference, recurrence risks, and implications for disease follow-up should be considered when deciding whether to treat with RAI (8).

Thyroid hormone replacement

Optimal TSH goal

Advocates of thyroid hormone suppressive therapy suggest that it can prevent the growth and spread of DTC (143, 144). Therefore, thyroid-stimulating hormone (TSH) suppression with l-thyroxine (LT4) is a therapeutic option for patients who undergo thyroid surgery. In 2013, when Haymart et al. (24) evaluated physician practice preferences, they found that there was consensus between endocrinologists on the role of TSH suppression for high-risk DTC but not for low-risk DTC. For a 0.8-cm PTC, 254 (50%) of physicians would suppress TSH. Similarly, 300 (56.5%) would suppress TSH for a 1.1- to 2-cm PTC. Because there was close to a 50%/50% divide on the role of TSH suppression for low-risk PTC, this study suggested a lack of consensus and a tendency for more aggressive treatment than guideline recommendations. In contrast, there was more physician agreement on the role of TSH suppression if lymph node or distant metastases were present, as 465 (91.9%) and 493 (97.2%) would suppress TSH, respectively.

Absorption of LT4 occurs in the jejunum and ileum (145). Coadministration of food with LT4 is likely to impair LT4 absorption; therefore, when possible, LT4 should be taken consistently either 60 minutes before breakfast or at bedtime (3 or more hours after the evening meal) for optimal and consistent absorption (146). Use of different LT4 products may alter serum thyrotropin (TSH) values in the same individual. Therefore, a change in an identifiable formulation of LT4 (brand name or generic) should be followed by re-evaluation of serum TSH at steady state (146). The recommendation for maintaining a particular identifiable formulation of LT4 is based on concern that products judged to be bioequivalent do not always have therapeutic equivalence and that switching products could lead to fluctuations in serum TSH (146, 147). This is especially a concern in frail patients, those with high-risk thyroid cancer, children, and pregnant women (146).

Thyroid hormone supplementation should be initiated to treat postsurgical hypothyroidism. Compared with the prethyroidectomy levels, higher serum free T4 levels are present after total thyroidectomy in patients receiving LT4 treatment and especially among those with suppressed TSH (148). In patients who undergo thyroidectomy, higher serum T4 levels are necessary to obtain serum T3 concentrations similar to those of euthyroid patients with an intact thyroid gland because of the need to compensate for the absence of the 20% fraction of circulating T3 secreted directly by the thyroid gland (148–151).

Although all patients need thyroid hormone replacement after total thyroidectomy, thyroid hormone supplementation may not be needed after thyroid lobectomy if patients can maintain their serum TSH within the target range (8). A meta-analysis of 50,445 patients, including 15,412 (30.6%) who underwent lobectomy, found that the incidence of postoperative hypothyroidism after lobectomy was 10.9% to 48.8% (152). Higher preoperative TSH levels (>2.5 mIU/L), presence of antithyroid antibodies, and presence of Hashimoto thyroiditis on surgical pathology predicted postoperative need for thyroid hormone supplementation (31, 153–155).

Therapy with suppressive doses of LT4 has long been known to positively affect outcome in DTC, and historic data suggest that an undetectable TSH is associated with longer relapse-free survival (40, 156). Until very recently, many patients with low-risk DTC would have been followed with suppressive doses of LT4 for 5 years after total thyroidectomy (157). The rationale had been that DTC expresses the TSH receptor on the cell membrane and responds to TSH stimulation. Therefore, measurable TSH levels could increase expression of thyroglobulin and increase rates of cell growth (158–160).

However, the traditional treatment paradigm was challenged in 1998 when the NTCTCGS registry demonstrated that TSH suppression to a low level reduced recurrence among patients with AJCC stage III and IV disease but was not beneficial in low-risk patients. In this registry, patients with stage I and II disease were considered to be “low risk,” and patients with stage III and IV disease were considered to be “high risk” (161). In 2006, a follow-up prospective study from the NTCTCGS registry demonstrated that TSH suppression was associated with improved OS in stage II patients with subnormal TSH levels compared with those patients with normal or elevated TSH levels [relative risk (RR), 101; 95% CI, 10 to 2300], whereas no significant difference was noted for stage I patients (RR, 6 × 10^–6; 95% CI approaches 0). For both stage I and II patients, there was no significant difference in DSS (RR, 1.00; 95% CI approaches 0 and RR, 1.6 × 10^–6; 95% CI approaches 0, respectively) or DFS (RR, 0.47; 95% CI, 0.077 to 1.54 and RR, 1.03; 95% CI, 0.17 to 3.43, respectively) (96). In a follow-up study of 4941 patients included in the NTCTCGS registry, maintaining TSH in the subnormal-normal range was associated with significantly improved OS (RR, 0.13 and 0.09 for stage I and II, respectively) and DFS (RR, 0.52 and 0.40 for stage I and II, respectively) (97). There was no survival benefit to more intensive TSH suppression.

Hovens et al. (162) assessed death and recurrence among 366 patients treated with DTC on LT4 after total thyroidectomy and radioiodine ablation. This was a single-institution study with a median follow-up of 8.85 years. Of the 366 patients, 310 patients had at least four unstimulated TSH measurements available. After 1 year of therapy, 250 (81%) were cured. Recurrence occurred in 39 (16%) of the “cured” patients, and 10 (4%) died secondary to thyroid cancer. Median TSH levels >2 mU/L were associated with thyroid cancer–specific death (HR, 2.03; 95% CI, 1.22 to 3.37) and recurrence (HR, 1.41; 95% CI, 1.03 to 1.95), whereas there were no differences in the rates of deaths and recurrences among patients with serum TSH levels between 0.1 and 0.4 mU/L (162). Collectively, these results from NTCTCGS and Hovens et al. suggest that less aggressive TSH suppression is reasonable for low-risk patients (163).

Informed by the recent data, the 2015 ATA guidelines for DTC have liberalized TSH goals (8). For low-risk patients with undetectable serum thyroglobulin levels after total thyroidectomy, TSH may be maintained at the lower end of the reference range (0.5 to 2.0 mU/L). For low-risk patients who have undergone remnant ablation and have low serum thyroglobulin levels, TSH may be maintained at or slightly below the lower limit of normal (0.1 to 0.5 mU/L), while surveillance for recurrence is continued. Similar recommendations are applicable for low-risk patients who have not undergone remnant ablation but have a low level of detectable thyroglobulin, although serum thyroglobulin levels may be measurably higher. Although there are little data on optimal TSH goals post lobectomy, per the 2015 ATA guidelines, for low-risk patients who have undergone lobectomy, TSH may be maintained in the mid to lower reference range (0.5 to 2 mU/L), while surveillance for recurrence is continued. After lobectomy, thyroid hormone supplementation may not be needed if patients can maintain their serum TSH in this target range (8).

Balancing benefits-risks and addressing limitations of available data

When determining the potential role of thyroid hormone suppression as treatment, both study limitations and patient risks should be considered. Unfortunately, national cancer registries such as SEER and NCDB do not capture thyroid function tests, and therefore many of the studies evaluating thyroid hormone suppression have smaller cohorts than the studies evaluating thyroid surgery. In addition, a patient’s TSH level can fluctuate, and longitudinal data are often not available.

For most patients with low-risk DTC, the goal TSH level is in the low-normal range. However, in select patients where TSH suppression may be warranted, the benefit needs to be balanced against risks. The primary risks are related to the heart and bone. Older patients are especially vulnerable to these risks. Adverse cardiac effects of TSH suppression include increased risk for atrial fibrillation among patients >60 years and increased risk of cardiovascular and all-cause mortality (164, 165). Flynn et al. (166) performed a population-based study of all patients in Tayside, Scotland, on LT4 (N = 17,684 patients). Patients were categorized by TSH, with suppressed TSH ≤0.03 mIU/L, low TSH 0.04 to 0.4 mIU/L, normal TSH 0.4 to 4.0 mIU/L, and raised TSH >4.0 mIU/L. With a median follow-up of 4.5 years, patients with suppressed TSH (≤0.03 mU/L) were at increased risk for cardiovascular mortality (HR, 1.37; 95% CI, 1.17 to 1.60), dysrhythmias (HR, 1.6; 95% CI, 1.10 to 2.33), and fractures (HR, 2.02; 95% CI, 1.55 to 2.62) (166). Although the proportion in this study with hypothyroidism secondary to treatment of thyroid cancer is unclear, Klein Hesselink et al. (164) compared 524 patients with DTC vs 1572 sex- and age-matched controls. This study showed that the risk of cardiovascular mortality is 3.3-fold increased and the risk of all-cause mortality is 4.4-fold increased among patients with DTC compared with controls, independent of age, sex, and other cardiovascular risk factors. Each 10-fold decrease in geometric mean TSH level is independently associated with a threefold increased risk of cardiovascular mortality (164). These data support tempering TSH suppression among patients with DTC at low risk of cancer recurrence. In addition, a recent retrospective study of 87 patients aged >75 years with DTC status post thyroid surgery and on LT4 showed similar recurrence rates at 5-year follow-up between patients with TSH values <0.1 mIU/L, 0.1 to 0.3 mIU/L, or 0.3 to 0.5 mIU/L. However, arrhythmias (P = 0.019), osteoporosis (P < 0.001), and anxiety/insomnia (P = 0.008) were significantly increased in the group with lower TSH values (167).

Osteoporosis is a known risk of prolonged TSH suppression. A recent study by Wang et al. (168) evaluated risk of osteoporosis and recurrence among 771 patients with low- to intermediate-risk DTC on LT4 status post total thyroidectomy. The median follow-up was 6.5 years. Of the 771 patients, 29 (3.9%) were diagnosed with postoperative osteoporosis and 43 (5.6%) with DTC recurrence. Patients with median TSH levels ≤0.4 mIU/L had similar recurrence rates but more osteoporosis than those with TSH values >0.4 mIU/L (168). The increased risk of postoperative osteoporosis decreased when median TSH was maintained around 1 mIU/L (168). In a study by Kim et al. (169) in which the average TSH was 0.5 mIU/L in the first year following thyroidectomy, LT4 therapy was associated with decreased bone mineral density in both the hip and spine. The decline in bone density became insignificant after passing the initial 1-year postoperative period.

“TSH suppression is known to be associated with increased risk of osteoporosis among postmenopausal women”

TSH suppression is known to be associated with increased risk of osteoporosis among postmenopausal women (170). There is no clear evidence that TSH suppression leads to increased risk of osteoporosis in men, but cohort sizes have been small (171–173). In a longitudinal study conducted by Karner et al. (174) that included nine men with thyroid cancer, there was no change in bone mineral density except for loss of bone mass in the distal radius. Data are conflicting among premenopausal women; some studies have shown no significant change in bone mineral density, whereas others have demonstrated decreased bone mineral density (168, 172–179).

In conclusion, evidence to date suggests that the increased risk of osteoporosis in patients with suppressed TSH seems to be more pertinent for postmenopausal women rather than men and premenopausal women. Adverse cardiovascular effects are also more concerning in older patients with suppressed TSH levels as compared with younger patients. Clinicians should take into consideration the stage of the patient’s thyroid cancer, risk of recurrence, underlying comorbidities, and patient age when deciding the degree to which the serum TSH level should be suppressed. However, for most patients with low-risk DTC, TSH suppression is typically not warranted, and the goal TSH should be 0.5 to 2.0 mIU/L.

To further elucidate the goal TSH level in low-risk DTC, longitudinal studies with serial TSH values are necessary. These studies will be difficult to design secondary to the large sample size needed, the number of confounding variables, and the length of follow-up necessary. However, given the excellent prognosis of low-risk DTC and the risks of TSH suppression in select populations, in most scenarios, physicians and patients should avoid unnecessary TSH suppression for low-risk DTC until data suggest otherwise.

Long-term surveillance of low-risk DTC

Most of the recommendations for long-term surveillance of low-risk DTC are based on low-quality evidence. Prior work has shown that when clinical guidelines are based on expert opinion instead of strong evidence, there is often variation in care and lack of guideline concordance (180). Therefore, it is not unexpected that when 534 endocrinologists and nuclear medicine physicians were surveyed, there was a lack of consensus over all aspects of long-term surveillance for DTC (24). When physicians were asked which tests they routinely scheduled during the first year after thyroid surgery, 371 (70%) included random thyroglobulin, 196 (37%) stimulated thyroglobulin, 215 (41.4%) neck ultrasound, and 286 (54%) diagnostic whole-body RAI scan. When specifically asked about imaging during the first year after thyroid surgery in a patient with stage I DTC with thyroglobulin <0.5 ng/mL, 237 (45.4%) used neck ultrasound and 157 (29.6%) diagnostic whole-body RAI scan (24).

Among all patients with DTC, 20% to 30% will experience a recurrence, and 4% to 7% will die during the 10 years after their initial diagnosis (2, 6, 181, 182). However, for patients with low-risk DTC, the likelihood of recurrence is just 1% to 5%, and the likelihood of death from thyroid cancer is nearly 0% (6, 8). Therefore, the primary intent of surveillance for patients with low-risk DTC is to re–risk stratify, identify recurrence early, and tailor treatment/surveillance. Neck ultrasound and serum thyroglobulin analysis (unstimulated or stimulated) with thyroglobulin antibodies are recommended for re–risk stratifying patients and are the underpinnings of long-term follow-up (8, 183). In the section below, standard long-term follow-up in patients with low-risk DTC is reviewed, including benefits, risks, and limitations of current surveillance strategies.

Thyroglobulin

Thyroglobulin, a glycoprotein stored in the colloid of thyroid follicles, is a sensitive tumor marker, as detectable or increasing thyroglobulin levels are highly indicative of persistent or recurrent DTC (114). Thyroglobulin levels can be measured in an unstimulated fashion (while taking levothyroxine) or in a stimulated state (after thyroid hormone withdrawal or after administration of rhTSH). Postoperative (stimulated or unstimulated) thyroglobulin values are an important prognostic factor that can be used to guide clinical management. Given its half-life of 1 to 3 days, the postoperative thyroglobulin level should reach its nadir by 3 to 4 weeks postoperatively in nearly all patients (184–189). However, despite its strengths as a tumor marker, uncertainty exists over the optimal cutoff level for both stimulated and unstimulated postoperative thyroglobulin levels 4 to 6 weeks after surgery (8). Most studies to date have focused on the role of thyroglobulin levels post total thyroidectomy and RAI treatment. Although recent work has evaluated the role of serum thyroglobulin levels after total thyroidectomy without RAI treatment, there are still a paucity of studies evaluating interpretation of thyroglobulin after lobectomy.

In the follow-up of low-risk DTC, it is recommended to monitor serum thyroglobulin levels every 6 to 12 months during the first year of follow-up. In the absence of antibody interference, serum thyroglobulin levels have a high degree of sensitivity and specificity to detect DTC, especially after total thyroidectomy and remnant ablation. Serum thyroglobulin after thyroid hormone withdrawal is strongly related to residual tumor mass. Patients with higher serum thyroglobulin tend to have larger tumor masses (190). After ablation, an rhTSH stimulated thyroglobulin above 2 ng/mL is very unlikely if the unstimulated serum thyroglobulin is <0.1 ng/mL (191). Not only is postablation thyroglobulin useful, preablation thyroglobulin can also predict outcomes. In patients at low risk for recurrence, serum thyroglobulin measurements <10 ng/mL prior to remnant ablation may be useful for predicting subsequent disease-free status (192).

Simultaneous to measuring the thyroglobulin level, thyroglobulin antibody levels should be measured. Antithyroglobulin antibodies are present in 10% of the general population and in approximately 20% to 25% of patients with DTC (193). Antithyroglobulin antibodies will interfere with all assays for thyroglobulin, thereby making the measured thyroglobulin level unreliable. Although the thyroglobulin level is unreliable if antibodies are present, thyroglobulin antibodies can be used as an imprecise surrogate marker for residual thyroid remnant or persistent/recurrent disease. If thyroglobulin antibodies are measured with the same assay over time, the disappearance of antithyroglobulin antibodies over time suggests resolution of thyroid cancer cells in patients, and conversely, a rise in thyroglobulin antibodies with or without a rise in thyroglobulin suggests the possibility of recurrent disease (194, 195). Although there is limited evidence, there is consensus that rising thyroglobulin antibodies should prompt attempts to identify structural recurrence with ultrasound and/or other diagnostic procedures, such as radioiodine WBSs (195).

Several thyroglobulin assays are available, each with specific functional sensitivities. Immunometric assays are prone to interference from antithyroglobulin autoantibodies, which commonly cause falsely low serum thyroglobulin measurements. Moreover, variability in thyroglobulin autoantibody assays can result in a misleadingly undetectable serum thyroglobulin level secondary to antibodies that are present but not detected (196). There is no method that reliably eliminates thyroglobulin antibody interference, but radioimmunoassays for thyroglobulin may be less prone to antibody interference (197–199). Unfortunately, radioimmunoassays for thyroglobulin are not as widely available and may be less sensitive than immunometric assays in detecting small amounts of residual tumor, and their role in the clinical care of patients is uncertain. Liquid chromatography-tandem mass spectrometry (LC/MS-MS) assays of thyroglobulin may be a viable solution for measuring thyroglobulin in the setting of thyroglobulin antibodies. LC/MS-MS overcomes the issue of thyroglobulin antibody interference by using tryptic digestion of patient serum with subsequent measurement of thyroglobulin-proteotypic peptides by LC/MS-MS (200–202). Additional outcome studies and studies correlating mass spectrometry assay results with immunoassays are needed (200, 202, 203). Most of the published data available on thyroglobulin come from studies that have a thyroglobulin assay sensitivity of 1 ng/mL. However, the sensitivity of most of the current assays is ≤0.1 ng/mL (204, 205).

Most studies have evaluated the role of thyroglobulin levels after total thyroidectomy followed by RAI. Malandrino et al. (206) found that in a series of 425 patients with DTC status post total thyroidectomy and RAI, a basal thyroglobulin ≤0.15 ng/mL was associated with low likelihood of residual disease. Brassard et al. (207) evaluated recurrence in 715 patients with DTC status post total thyroidectomy and lymph node dissection. In this cohort, 94% were treated with RAI for remnant ablation. During the median follow-up of 6.2 years, 32 patients had recurrence. A thyroglobulin cutoff of ≤0.27 ng/mL had maximal sensitivity (72%) and specificity (86%) for recurrence. With this thyroglobulin cutoff, 117 (16%) had a positive thyroglobulin level, and of these, 23 (20%) had recurrences. Recurrence was observed among just 9 (1.5%) of the 598 patients with thyroglobulin levels below this cutoff (207).

The recent 2015 ATA guidelines emphasize the importance of response to therapy reclassification in patients with DTC treated by total thyroidectomy and RAI (8). Response to therapy can be categorized as “excellent response, biochemical incomplete response, structural incomplete response, and indeterminate response” (6, 8). For an excellent response, a patient has negative imaging and suppressed thyroglobulin <0.2 ng/mL or stimulated thyroglobulin <1 ng/mL. Biochemical incomplete response includes negative imaging but nonstimulated thyroglobulin that is ≥1 ng/mL, stimulated thyroglobulin that is ≥10 ng/mL, or thyroglobulin antibodies that are rising. Structural incomplete response is any structural disease. Indeterminate response includes nonspecific imaging findings, faint uptake in the thyroid bed on RAI scanning, nonstimulated thyroglobulin levels that are detectable but <1 ng/mL, stimulated thyroglobulin that is detectable but <10 ng/mL, or antithyroglobulin antibodies that are stable or declining (8). Per the recent 2015 ATA guidelines, patients are deemed to be disease free if, status post total thyroidectomy and RAI, their thyroglobulin levels during LT4 supplementation are <0.2 ng/mL or after stimulation are <1 ng/mL, assuming no antibodies and no imaging evidence of disease. In this scenario, thyroglobulin testing can be spaced out to every 12 to 24 months (8). In fact, if the thyroglobulin level is low with a sensitive thyroglobulin assay, then obtaining a stimulated thyroglobulin level may be unnecessary (207).

Recently, studies have evaluated the role of thyroglobulin levels over time in patients who did not undergo RAI remnant ablation after total thyroidectomy. Using a sensitive assay to confirm the utility of thyroglobulin measurements on thyroid hormone treatment of routine follow-up, Durante et al. (208) demonstrated that in 290 low-risk DTC patients who had not undergone remnant ablation, serum thyroglobulin levels on LT4 therapy became undetectable (<1 ng/mL) within 5 years in 95%. The levels were <0.2 ng/mL in 80% of these patients. In a retrospective review of 76 consecutive low-risk DTC patients treated with total thyroidectomy alone and followed for a median of 2.5 years, an unstimulated thyroglobulin of ≥2 ng/mL with a concomitant median TSH level of 0.48 mIU/L was reported to be sufficient to detect all patients with disease recurrence (108). In ATA low-risk patients who did not receive RAI remnant ablation, an unstimulated postoperative thyroglobulin level <1 ng/mL was associated with excellent clinical outcomes and recurrence rates <1% with a median follow-up of 62 months (102).

Although the specific thyroglobulin cutoff for disease-free status is unclear in patients status post total thyroidectomy but no RAI, a rising thyroglobulin level over time is suspicious for growing thyroid tissue or thyroid cancer recurrence. In patients who have not received RAI and who fall into the ATA low-risk category, there is no need for rhTSH stimulation because thyroglobulin levels will increase to values >1 ng/mL in 50% of cases, and even in individuals without residual cancer. The magnitude of thyroglobulin increase with rhTSH stimulation is related to the size of the normal thyroid remnants (209, 210).

The value of obtaining thyroglobulin levels post thyroid lobectomy is even less clear (211). Although agreed that a marked rise in thyroglobulin over time is suspicious for recurrence, the ideal cutoff is unknown. In 2016, Momesso et al. (212) studied response to therapy in 187 patients with DTC treated with lobectomy and 320 status post total thyroidectomy without RAI. Of these patients, 85.4% had low-risk disease; median follow-up was 100.5 months. None of the patients experienced recurrence if unstimulated thyroglobulin levels after total thyroidectomy were <0.2 ng/mL and after lobectomy were <30 ng/mL, assuming undetectable thyroglobulin antibodies and negative imaging. More studies on thyroglobulin trends after lobectomy are needed, because thyroglobulin level is closely related to remnant thyroid and TSH level (212).

Imaging

Cervical lymph nodes metastases are the most common site of recurrence for DTC, and especially PTC. Several imaging studies are available to identify recurrent thyroid cancer, but standard surveillance imaging typically includes obtaining a neck ultrasound at 6 to 12 months post thyroid surgery. During long-term surveillance, diagnostic whole-body scanning with radioiodine, magnetic resonance imaging, computed tomography, and fludeoxyglucose-positron emission tomography scanning are usually not indicated in low-risk patients; rather, neck ultrasound is the most sensitive imaging technique for the detection of metastatic cervical lymph nodes as well as recurrences or residual thyroid cancer in the thyroidectomy bed (213–215).

When performing a neck ultrasound, it is important that the thyroid bed and all compartments of the anterior and lateral neck are evaluated. The ultrasound characteristics suspicious for malignancy in lymph nodes include cystic appearance, microcalcifications, loss of the normal fatty hilum, and peripheral vascularization. As a single criterion, round shape, hypoechogenicity, or the loss of a fatty hilum is not specific enough to indicate malignancy (214). If worrisome lymph nodes that are ≥8 to 10 mm in their smallest diameter are identified, ultrasound-guided FNA should be performed prior to consideration of a neck dissection. Ultrasound guidance improves the results of FNA biopsy for small lymph nodes and those located deep in the neck. However, because FNA cytology can miss thyroid cancer in up to 20% of patients, combining cytology and thyroglobulin levels in the aspirate fluid increases sensitivity (216–218). In cases of lymph node metastases, the thyroglobulin concentration in the aspirate fluid is often elevated, with concentrations >10 ng/mL associated with a high level of suspicion for lymph node metastasis (219–221). An aspirate thyroglobulin level between 1 and 10 ng/mL is moderately suspicious for malignancy, and a comparison of the thyroglobulin measurement in the aspirate fluid to the serum should be considered in these patients. If lymph nodes are consistent with recurrent thyroid cancer but no surgical intervention is recommended, these lymph nodes can be monitored for progression by serial neck ultrasonography. In addition, small lymph nodes (<8 to 10 mm in largest diameter) and lymph nodes that are not so worrisome for metastatic disease can be monitored with neck ultrasound, especially because reoperation is associated with elevated risk (8).

Although there are many benefits to ultrasound surveillance and ultrasound-guided FNA, it is important to maintain a broad differential diagnosis when performing ultrasound exams, particularly in thyroid cancer patients with low or undetectable thyroglobulin levels (213). In ATA low-risk patients without structural evidence of disease on initial surveillance evaluation, routine screening with ultrasound is more likely to identify false-positive results than clinically significant structural disease recurrence. Yang et al. (222) followed 171 patients with low-risk disease for a median of 8 years. These patients underwent a median of five neck ultrasound (range, 2 to 17). Structural recurrence with disease ≤1 cm was identified in two (1.2%) patients. False-positive ultrasound findings occurred in 114 (67%) patients. Because up to one-half of the FNA biopsies performed for suspicious ultrasound findings are benign, selection of patients for FNA biopsy needs to be improved (220, 221, 223). Therefore, use of ultrasound and ultrasound-guided FNA should be tailored to the clinical suspicion of recurrence. If the initial ultrasound 6 to 12 months post surgery is reassuring, then repeating it is not indicated unless there is a measurable or rising thyroglobulin level or if there is a palpable lymph node in the neck (8).

Although diagnostic WBSs had been part of standard care in the past, they are no longer routinely recommended for long-term follow-up of low-risk DTC (142). The indications for when to use diagnostic WBSs remain controversial. In patients in whom serum thyroglobulin levels remain elevated over time, and especially for those with rising thyroglobulin levels, if neck ultrasound is not revealing, then a diagnostic WBS can be considered. If remnant ablation or adjuvant therapy with I-131 is necessary, then a corresponding post-therapy WBS should be performed. However, there is no evidence in these low-risk patients that delayed treatment with I-131 adversely affects patient outcome (8). Thus, I-131 scans are not routinely recommended for low-risk DTC patients, but they can be considered when there is clinical suspicion of residual or recurrent iodine-avid disease.

Length and interval of follow-up

The optimal length and interval of long-term surveillance for low-risk DTC remains unknown. After surgery for low-risk DTC, patients typically have levels of thyroglobulin and thyroglobulin antibodies checked at around 3 to 6 months, after which they should be seen for physical examination and neck ultrasound. If physical examination, thyroglobulin and thyroglobulin antibody levels, and neck ultrasound are unremarkable, the patient can be followed yearly with unstimulated thyroglobulin levels with concurrent measurement of thyroglobulin antibodies (8). However, in a low-risk patient, when to transition from yearly follow-up to longer intervals is unknown. In addition, the decision about when to transition follow-up from endocrinologists to primary care physicians remains an evolving topic.

Most thyroid cancer recurrences occur in the cervical lymph nodes within 5 years of initial treatment, but late recurrences occurring 10 or more years after the initial treatment have been described (224, 225). For this reason, some have suggested that even patients with low-risk DTC require lifelong surveillance (226). Although an evolving topic, recent data suggest that for younger patients (aged <45 years at diagnosis) with stage I disease, there is no proven survival benefit from life-long surveillance. After 5 years of follow-up, these patients could be safely discharged to primary care, with annual thyroglobulin level measurement thereafter (207, 227–233). In patients with low-risk PTC, negative neck ultrasound, and stimulated thyroglobulin <2 ng/mL 1 year postoperatively, Wu et al. (234) evaluated the cost-effectiveness of tapering postoperative surveillance to 3-year intervals after 5 years of annual surveillance. Probabilistic sensitivity analysis found that less frequent surveillance was more cost-effective in 99.98% of 10,000 simulated patients. Wang et al. (235) evaluated the cost of surveillance in 1087 patients status post thyroidectomy for DTC. The cost of surveillance for each recurrence detected was $147,819 in low-risk patients. This cost per recurrence in low-risk patients was 6 and 7 times higher than the cost of surveillance in intermediate- and high-risk patients, respectively, suggesting less intensive surveillance strategies are needed for low-risk DTC.

Balancing benefits-risks and addressing limitations of available data

The benefit of long-term surveillance includes re–risk stratifying, detecting, and treating potentially life-threatening recurrence. Risks include identifying clinically insignificant recurrence that may lead to unnecessary treatment and associated patient risks (236). As our thyroglobulin assays and ultrasounds are increasingly sensitive, we have the ability to detect a low level of residual or recurrent disease. Some of this disease may never be life threatening. Yet reoperation and additional RAI may be associated with patient harm (236–238). In fact, recent data show that we are detecting and treating more thyroid cancer recurrences, but in most scenarios, there is no improvement in DSS (181). Although periodic ultrasound is a key element to surveillance, diagnosing recurrence and detecting false-positive findings can lead to patient worry (239). Thus, another risk of long-term surveillance of patients with low-risk DTC is unnecessary patient distress due to false-positive findings and detection of clinically insignificant recurrences. Related to this, the optimal frequency of imaging and testing may need to be redefined.

Finally, the pendulum only recently swung toward less intensive treatment of low-risk DTC; therefore, there has not been enough time to assess how this practice change will impact long-term surveillance. Additional long-term outcomes data are needed prior to definitively defining optimal surveillance in an era of “less is more.”

Although long-term surveillance is an important element in the care of patients with low-risk DTC, the evolving treatment paradigms, low event rate, potential for false-positive findings with imaging, and subsequent patient worry make this a topic for which further research is needed. Perhaps more than any other domain, there is a need for additional studies on optimal long-term surveillance. Implementing these studies will take time, as the pendulum has only recently switched to “less is more,” and the length of follow-up necessary to compare surveillance strategies may be as long as 5 to 10 years. In addition to future studies, to avoid excessive imaging and surveillance post primary treatment of low-risk DTC, patient worry and anxiety should be addressed. Tools to reduce worry, both for the patient and physician, may help streamline surveillance strategies and reduce unnecessary and excessive surveillance.

“Cervical lymph node metastases are the most common site of recurrence for DTC”

Conclusion

For low-risk DTC, controversies exist across all phases of disease management. This controversy leads to variability in patient care, which can result in harm from overtreatment or undertreatment. To reduce controversy, increase physician consensus, and improve patient care, it is necessary to design rigorous studies to address current knowledge gaps and disseminate and educate both patients and physicians when quality data exist.