Introduction

Differentiated thyroid carcinoma (DTC) has an excellent prognosis with 10-year survival rates of 85–93% [1]. Total thyroidectomy (TT) followed by radioiodine remnant ablation (RA) is the usual treatment for DTC; nowadays the therapy is selected according to the risk of recurrence (RR) [2,3,4]. After this initial approach, serum thyroglobulin (Tg) level is an important marker for identifying those cases with persistent or recurrent disease [5,6,7]. However, the reliability of serum Tg measurement is significantly impaired by coexistent serum antithyroglobulin antibodies (sTgAb) [8,9,10,11]. This concept is so well established that the last American Thyroid Association guidelines for the management of DTC include the Tg and TgAb measurement after initial treatment as an essential tool to assess the ongoing risk stratification [5].

Antithyroglobulin antibodies (TgAb) are produced by immune cells, mostly by lymphocytes that react against thyroid auto antigens and progressively infiltrate the thyroid gland [12]. It is also well known that ~30% of DTC patients have serum TgAb which might interfere with Tg measurement [13, 14], mainly causing false negative results when Tg is measured with immunometric assays, and either, false positive or negative results, when it is measured with radioimmunoassay [15, 16]. In these cases, serum Tg loses its value as a tumor marker, but the change in the TgAb levels over time can be used as a surrogate marker, since it has been demonstrated that if TgAb levels decline and become undetectable, together with an undetectable Tg level, it will be indicative of the no evidence of disease status [17, 18]. On the other hand, the persistence of stable TgAb levels for a long time, more than 3 or 4 years, represents an alert and might indicate the possibility of persistence or recurrence [19, 20]. It has been observed that TgAb levels disappear gradually with an average of 3 years’ time after TT and RA, due to the removal of the Tg antigenic stimulus [18, 21, 22]. In contrast, there are few data about the change in TgAb levels over time in DTC patients treated with TT without RA.

Many studies have reported that the association between papillary thyroid cancer and Hashimoto’s thyroiditis (HT) might be an indicator of favorable prognosis but it is unclear whether the presence of TgAb per se has a significant impact in outcomes [23,24,25,26,27].

The aims of this study were: (i) to compare the outcome of DTC patients with detectable and undetectable TgAb levels stratified according to the initial RR, (ii) to compare the response to therapy in patients with detectable TgAb treated with TT with and without RA, and (iii) to evaluate the dynamics of the TgAb levels trend in the long-term follow-up according to the initial response to therapy.

Patients and methods

Data source and study population

We retrospectively reviewed our database containing 549 files records of patients with DTC who were followed-up from January 2015 to February 2018 in the Division of Endocrinology, Hospital de Clinicas—University of Buenos Aires. Inclusion criteria were: (i) age older than 18 years, (ii) detectable serum TgAb levels at some point during the follow-up, (iii) adequate clinical and pathological data to allow an accurate determination of the initial RR, (iv) a minimum follow-up of 12 months after initial treatment with, at least two Tg/TgAb assessments, to define the initial response to therapy. Eighty-seven patients were excluded due to lack of enough time of follow-up (less than 12 months) and 30 patients due to insufficient data in the follow-up. One hundred and six (24.5%) patients of the remaining 432 had detectable serum TgAb levels. The initial treatment was TT for all patients, with or without the concomitant administration of radioiodine RA.

Each patient was stratified by using the eighth edition of the American Joint Committee on Cancer/International Union against Cancer (AJCC/UICC) staging system and the RR was assessed by using the modified risk stratification system from the 2009 ATA guidelines proposed by the American Thyroid Association (ATA) (low, intermediate, and high) [5, 28].

Pathologically lymphocytic thyroiditis

Pathologically proven lymphocytic thyroiditis was defined as the presence of diffuse lymphocytic and plasma cell infiltrate and oxyphilic cells, and the formation of lymphoid follicles and reactive germinal centers. The infiltration had to occur in a normal region of the thyroid gland, distinct from the site of DTC. A peritumoral inflammatory response was not considered to be pathologically proven thyroiditis.

Ablation protocol

The decision of RA with radioiodine was taken according to the usual behaviors established historically by the ATA guidelines [5, 29]. Two hundred and ninety-six patients received RA: 35 after stimulation with recombinant human thyrotropin (rhTSH) and 261 after thyroid hormone withdrawal for at least 3 weeks. Radioiodine was administered in all cases with TSH levels above 50 mIU/L. Administered radioiodine doses of 131I ranged from 1.11–5.55 GBq (30–150 mCi 131I). One hundred and thirty-six patients did not receive RA after surgery.

Serum thyroglobulin and antithyroglobulin antibody (TgAb) measurement

Serum Tg and TgAb were assessed in one of two reference laboratories from Argentina, the same laboratory and assay were used throughout patient’s follow-up. Tg assay comprised the Elecsys Tg Electrochemiluminescence Immunoassay (Roche Diagnostics GmbH, Mannheim, Germany). The functional sensitivity was 0.1 ng/ml. TgAb assay comprised the Elecsys Anti-Tg Electrochemiluminescence Immunoassay (RSR Ltd, Pentwyn, Cardiff, UK), values > 20 IU/ml were considered positive, in accord with the manufacturerʼs recommendations.

Dynamics of change in TgAb levels

In patients who had at least two assessments of Tg and TgAb levels during the first year of follow-up, the trend of TgAb levels over time was calculated according to the following definitions:

  • Increase: elevation of TgAb levels more than 50%.

  • Stability: fluctuating TgAb levels or either elevation or decrease less than 50%.

  • Decrease: decrease of TgAb levels more than 50%.

  • Undetectability: at some point during the follow-up.

Response to therapy

The initial response to therapy were assessed based on serum Tg and TgAb measurements, neck ultrasonography (US), diagnostic whole-body scans (dxWBS), and additional morphological/functional images when appropriate. Patients were considered to have an excellent response (ER) to therapy if they had undetectable suppressed Tg values or a stimulated Tg < 1 ng/ml, absence of TgAb, no uptake outside thyroid bed dxWBS (if done), and no suspicious lymph or thyroid bed nodules on neck US. Patients with stimulated serum Tg between 1 and 10 ng/mL or nonstimulated Tg values between 0.2 and 1 ng/mL and no structural evidence of disease or nonspecific findings in the neck US or other cross-sectional images, or with persistent measurable TgAb, were considered as having an indeterminate response (IR). Patients with stimulated serum Tg > 10 ng/mL or nonstimulated Tg levels > 1 ng/mL, or those with increasing TgAb without structural evidence of disease, were classified as having a biochemical incomplete response. Patients with structural incomplete response (SIR) were those showing positive cytology or histology, highly suspicious findings on neck US or findings on dxWBS, and functional or morphological imaging suggesting DTC metastasis.

According to the clinical status at the end of follow-up, patients were classified as having no evidence of disease (NED) if they had undetectable suppressed Tg values or a stimulated Tg < 1 ng/ml, undetectable TgAb, ultrasonographically no suspicious findings, and no evidence of structural disease on any other imaging studies. The remaining group of patients was classified as having an indeterminate, biochemical, or SIR by using the same definitions that were described above. Recurrent disease was defined as having structural or biochemical evidence of disease following a period of NED.

Clinical management during follow-up

Each patient was assessed with Tg and TgAb measurements under hormonal therapy, and neck US every 6 months after initial treatment. Thyroid US was performed using a 13-MHz linear transducer. Central and bilateral neck lymph node compartments and the superior mediastinum were inspected. Suspected lesions were evaluated by US-guided fine needle aspiration cytology (FNAC) and measurement of Tg and TgAb in washing fluid.

In patients with a biochemical incomplete response during the follow-up, morphological or functional imaging was performed. These studies included US, computed tomography (CT), magnetic resonance imaging (MRI), and/or nuclear imaging procedures such as whole-body scans after a diagnostic or therapeutic radioiodine dose or the use of positron emission tomography/CT using fluorine-18 fluorodeoxyglucose (18F-FDG-PET/CT scan). After initial therapy, all patients received hormonal therapy to keep a TSH level according to the RR and response to therapy during the follow-up according to the ATA guidelines.

Statistical analysis

Epidemiological data are presented as the mean ± SEM, with median and range when appropriate. For categorical variables, the number and percent of patients and/or scans within each category was determined. The categorical variables were compared by Pearson Chi-square and the continuous variables by the Mann–Whitney U test. A p value < 0.05 was considered statistically significant. All statistical operations were performed using the Stata 14.1 (Stata Corp, TX, and USA). The Kaplan–Meier method with the log-rank test was used to analyze time to undetectable TgAb. Hazard ratios (HR) and confidence intervals using log-rank analysis were also calculated. Univariate and multivariate cox proportional hazards model were used using time to NED as the outcome variable these results were expressed as the HR with a 95% confidence interval. The initial risk was considered as a covariate.

Results

Patientʼs characteristics

The demographic and clinical features of the 432 patients included in this study can be observed in Table 1. The majority had the classic variant of papillary thyroid carcinoma (94.2%); they were female (81.7%) and were AJCC stage I (81.3%). According to the ATA RR classification, patients were considered as having low (53.7%), intermediate (26.8%), or high RR (19.4%). The median follow-up in the whole cohort was 53 months (range: 12–354; mean 51 ± 41). There were no statistically significant differences when baseline characteristics were compared in patients with negative and positive TgAb, except for the presence of lymphocytic thyroiditis, as expected (Table 1).

Table 1 Baseline characteristics of 432 patients with differentiated thyroid cancer included in the study
Full size table

Response to treatment and status at final follow-up in patients with negative and positive antithyroglobulin antibodies according to the initial risk stratification system of the American Thyroid Association

The frequency of SIR after the initial treatment was 6.5% (n = 15), 20% (n = 23), and 69% (n = 58) in low, intermediate, and high RR, respectively. There were no statistically significant differences between patients with positive and negative TgAb (Table 2a–c).

Table 2 Response to treatment and status at final follow-up in patients with negative and positive antithyroglobulin antibodies
Full size table

Patients with low, intermediate, and high RR with positive and negative TgAb had similar frequency of SIR at the end of follow-up (Table 2a–c). As expected, patients with detectable TgAb classified as having low and intermediate RR, had less frequency of ER and higher percentages of IR at the initial evaluation and at the end of follow-up (Table 2a, b). The percentage of distant metastases and additional treatments were similar in patients with positive and negative TgAb in all groups of recurrence (Table 2a–c).

Antithyroglobulin antibodies trend and radioiodine RA

Low-risk patients treated with or without RA achieved the same frequencies of SIR or excellent responses both initially and at 36 months of follow-up. As the median follow-up of patients treated without RA was 27 months (range 16–42), it was not possible to compare the responses between these groups beyond 36 months of follow-up (Table 3).

Table 3 Baseline characteristics and comparison of the responses to therapy in patients with detectable TgAb with and without radioiodine remnant ablation
Full size table

Undetectable TgAb levels in patients treated with RA were achieved in a median of 16 months (range 11–62) vs. 11 months (range 6–21) in patients treated without RA (p = 0.0232). The trend of TgAb levels in the first 12 months was clearly different: non-ablated patients had a higher rate of undetectable levels while patients who received RA showed a more stable TgAb concentration during follow-up (Fig. 1).

Fig. 1
figure 1

Kaplan–Meier curve: time to undetectable antithyroglobulin antibodies in low-risk patients with and without radioiodine remnant ablation

Full size image

Antithyroglobulin antibodies levels at first postoperative evaluation

The median initial TgAb levels were 127 IU/ml (range: 0–3320). According to the RR, the values were: 116 IU/ml (range: 0–2000) in low RR patients, 182 IU/ml (range: 25–3320) in intermediate RR patients, and 96 IU/ml (range: 0–640) in high RR patients (p = 0.617). The initial mean levels in patients with SIR were 480 IU/ml (±420 IU/ml) and in patients without SIR 280 IU/mL (±36 IU/ml) (p = 0.098). The initial TgAb levels did not correlate with the presence of SIR at the initial assessment. In the ROC curve we observed that there was not a specific TgAb level which might predict the presence of SIR to treatment (Fig. 2).

Fig. 2
figure 2

Correlation of antithyroglobulin antibodies levels at first postoperative evaluation in patients with structural incomplete response during follow-up

Full size image

Antithyroglobulin antibodies trend during the first year of follow-up in patients without apparent clinical disease (n = 70)

Patients with an initial structural incomplete disease were excluded for this analysis. A SIR at the end of follow-up was observed in the 3% of patients with decreasing TgAb levels, in the 19% with stable levels, and in the 43% of patients with increasing levels. Therefore, the status of no evidence of disease was achieved in 51.5% of patients with decreasing TgAb levels, in 31% of those with stable values while in none with increasing levels. The optimal cutoff value of the TgAb at 12 months of follow-up to predict a SIR as final status was established at 160 IU/ml, with a sensitivity (S) of 75% and a specificity (E) of 71.9%. A TgAb level lower than 38 IU/ml predicted an undetectable outcome at some point of the follow-up with an S of 90% and an E of 74%. Consequently, the TgAb trend was more useful than an absolute value of TgAb at the initial assessment or at the first year of follow-up to predict the final outcome.

Discussion

The incidence of TgAb is approximately twofold higher in DTC patients, especially in papillary thyroid carcinoma [30, 31], compared with the general population (20% vs. 10%, respectively), suggesting an association between autoimmune thyroid disease and DTC [8, 13, 21, 32,33,34]. Indeed, many studies reported an association between HT and DTC [14, 31, 35,36,37].

There were controversial reports about the relationship between lymphocytic thyroiditis and anti-TgAb with the outcome in DTC patients [23,24,25,26,27]. In general, lymphocytic thyroiditis is considered a favorable prognostic factor for papillary thyroid cancer (longer disease-free survival and decreased mortality), although this has not always achieved statistical significance using multivariate analysis [23,24,25,26, 31, 38, 39]. Lymphocytic thyroiditis is highly correlated with the serological presence of thyroid antibodies; however, whether the presence of TgAb per se has prognostic significance remains controversial [14, 21, 36, 40, 41]. Our study compared the outcome of DTC patients with detectable and undetectable TgAb levels, taking into account not only the initial RR, but also the response to therapy. The frequency of SIR in patients with detectable and undetectable TgAb levels of low, intermediate, and high RR was similar, not reaching statistical significant differences.

In the last years, the use of radioiodine RA has been reconsidered, and currently, a rather large number of DTC patients are not treated with radioiodine [5]. In these cases, the change of the serum Tg levels over the time, more than its absolute value (undetectable or detectable), still represents a strong tumor marker [6, 7]. At the same time, the persistence of stable levels of TgAb for a long period or the increase of TgAb levels after TT and RA represent an alert indicating the possibility of persistence or recurrence of the disease [19, 20].

While the use of TgAb trend as a surrogate marker has been recognized as a valuable tool in patients treated with TT and RA [19], few data has been reported about the changes of TgAb levels in patients treated with TT without RA [42, 43]. The disappearance of the antigenic stimulus (i.e., the absence of any thyroid tissue) induces a progressive reduction and ultimately a disappearance of thyroid antibody production, although the survival of a substantial fraction of plasma cells can be responsible of a temporary persistence of humoral immunity [44]. The question of whether the presence of a normal postoperative thyroid remnant could maintain the stimulus for the antibody production, thus affecting the disappearance of the serum TgAb, has not been clarified yet. As a consequence, the positivity of TgAb levels in the follow-up of patients with DTC and positive TgAb treated with TT but without RA represents a matter of concern. We compared the response to therapy in low-risk patients with detectable TgAb levels treated with TT with and without RA; both groups achieved the same initial and final response to treatment. However, TgAb became negative in a longer period in patients treated with RA and the trend of TgAb levels in the first 12 months was significantly different: more patients treated without RA had undetectable TgAb levels while those ablated patients had a higher proportion of stable TgAb levels during this period. This situation was also reported in a previous study [20]. Transient increases or stable levels in TgAb concentration can occur after 131I therapy and a prolonged period may be necessary for their return to baseline concentrations [18, 19]. It is now clear that removing the Tg antigenic stimulus by TT and RA results in the eventual disappearance of TgAb over a median time of 3 years [18, 21, 22]. Conversely, TgAb concentrations may rise, or become detectable de novo, in response to an acute increase in Tg antigen after either the initial or secondary thyroid surgeries [45], FNA biopsy [46, 47], or radioiodine therapy [30, 48, 49]. More importantly, a high number of studies have reported that the de novo appearance, persistence, or a rising trend in TgAb concentrations in the postoperative period is a significant risk factor for persistent/recurrent disease [8, 19, 21, 32, 50, 51].

Accordingly, we found that the initial TgAb levels did not correlate with the presence of SIR and TgAb trend is more useful than the absolute value of TgAb at the initial assessment or at the year of the follow-up.

One of the strengths of our study was the evaluation of the dynamics of TgAb in patients treated without RA compared with those who had received it, and we observed that TgAb became undetectable earlier in the first group of patients. However, there are several limitations in our study, such as its retrospective design, the low number of patients in each of the individual groups when they were divided by the presence of TgAb and RR, in addition to the short follow-up of the cohort.

In summary, our study shows that the prognosis of DTC in patients with detectable and undetectable TgAb is similar, even after considering the initial RR. RA influences the trend of TgAb during the first year of follow-up, but not the response to therapy at the end of follow-up when the RR is considered. Our study confirms previous data regarding the usefulness of the TgAb trend, which can be surely used as a surrogate marker, and that it is more useful than an absolute TgAb value to predict a structural incomplete response to therapy.