**3. Discussion**

Extrathyroidal extension of thyroid cancer has been recognized as a factor of poor outcome [2,3]. More recent analyses have suggested that mETE is less significant for patient outcome than gross extension [5–7,18,19]. The updated eighth edition of the AJCC/TNM staging system removed the subclassification of mETE, resulting in a downstaging of T3 tumors ≤ 4 cm. However, the impact of mETE on treatment success after initial RAI therapy still remains a matter of debate. Although TNM-staging is mainly associated with the disease specific mortality risk, it still influences the clinical patient management. We therefore evaluated the primary presentation and the early treatment response in our patient cohort with or without mETE.

Firstly, we compared +mETE patients to a control group with matched histology, age, sex, and tumor size. Secondly, we subdivided patients in subgroups with N0/Nx- and N1a/N1b- stages to analyze the study group in more detail. These more homogenous subgroups were supposed to improve data quality, as confounding factors could be better analyzed.

Overall, our data suggested that mETE does not affect treatment success. Indeed, analyzing biochemical (Tg-levels) and structural (uptake in WBS and ultrasound) characteristics, comparable responder rates were found in both groups (+mETE vs. −mETE). These findings mirror data by Arora et al. and Kim et al., where the presence of mETE was not an independent predictive factor of recurrence [9,20]. Furthermore, the study by Ito et al. showed that mETE was not associated with recurrence in contrast to gross ETE. Thus, they suggested not using mETE as an indicator for poor prognosis [6].

Nevertheless, our data indicate that +mETE patients have a significantly higher risk of presenting with lymph node metastases at initial presentation as compared to −mETE patients. This is in line with the study of Kim et al. who reported that patients with gross or mETE were more likely to have nodal metastases in the central or the lateral neck compartments [20]. Shin et al. showed that there was no significant difference in recurrence between +mETE and −mETE patients; although lymph node metastases were an independent factor for the increased risk of mETE, it did not affect recurrence-free survival [7]. On the contrary, data in a recent meta-analysis from Diker-Cohen et al. indicated that, in patients with N1a/N1b disease, the presence of mETE did further increase the risk of recurrence, yet still within the low risk category [21]. Overall, our analysis identified N1a/N1b-stage as the only risk factor for lower responder rates in the multivariate analysis. Overall, our data indicate that patients with mETE have a higher risk of systemic disease. Therefore, since a prophylactic lymph node neck dissection is not suggested in patients with T1/T2 disease, precise perioperative diagnostic examinations (ultrasound, post-treatment WBS) are crucial in this cohort to exclude lymph node metastases [11].

In our study, response rates were similar in the N0/Nx- and the N1a/N1b-stage subgroup analysis. Nonetheless, it needs to be emphasized that, in this N-stage subgroup, +mETE patients received a significantly higher activity as compared to patients without mETE. Seo et al. compared the outcome after (adjuvant) initial RAI therapy in patients with lymph node metastases and mETE treated with standard activity (3.0 GBq) or a low-dose activity (1.1 GBq) and found no significant differences between both regiments regarding the initial response rate as well as during the follow-up within the first years (mean follow-up time of 45 months) [22]. Therefore, it is very unlikely that the significantly higher treatment activity in our +mETE group might have led to a significant bias in this study.

Patients with higher administered RAI activity had a worse treatment response compared to patients treated with lower RAI activity. This finding was to be expected, since significantly more patients with lymph node metastases were treated with higher doses. It is well known that patients with lymph node metastases show poorer treatment response [23].

It needs to be highlighted that patients included in this retrospective study come from a historical collective. Therefore, patients received relatively high doses of RAI, which was in line with the German and institutional guidelines at the time of treatment. However, current guidelines trend towards decreasing radioiodine doses, which was supported by the two prospective studies "Ablation with low-dose radioiodine and thyrotropin alfa in thyroid cancer" (HiLo) and "Strategies of radioiodine ablation in patients with low-risk thyroid cancer" (ESTIMABL) [24,25]. Compared to current guidelines, especially +mETE patients with lymph node metastases were most likely overtreated with RAI.

As the majority of PTC cases comprise classical and follicular variants of PTC, only these histological subgroups were included in our study [26]. Patients with classical PTC had mETE significantly more often as compared to patients with the follicular variant of PTC. This finding is in line with the study of Yu et al., who also reported significantly higher rates of ETE in classical PTC patients. Response rates in our patient cohort were comparable in both groups. Yu et al. also showed that no differences in outcome could be found between classical and follicular variants of PTC patients [27].

Our study has several limitations. First, there may have been a selection bias because of the retrospective design. Secondly, in our study, patients with mETE were treated with significantly higher radioiodine doses. Higher doses were administered because mETE was considered as a risk factor at the time of inclusion. Furthermore +mETE patients presented more often with lymph node metastases.

Thyroglobulin antibodies are only present in a minority of the patients and therefore could not be included in the analysis in a convincing way. This is indeed a limitation of the study, which cannot be overcome. However, although the recovery is less sensitive than the direct measurement of the antibodies, we are convinced that an undisturbed recovery adds confidence for the validity of the measured thyroglobulin.

Furthermore, primary surgical procedures were performed in different hospitals. Therefore, histological tissue or slides were not available, and only the original pathology reports were reviewed. However, we only included patients with detailed reports indicative for reclassification.
