Radiofrequency Ablation for Benign Thyroid Nodules: Ecacy According to the Types of Thyroid Nodules

Percutaneous radiofrequency ablation (RFA) has been recommended as minimally invasive treatment for patients with symptomatic benign thyroid nodules (BTNs) because of the large number of clinical applications. This retrospective observational study sought to evaluate the clinical outcomes of RFA for BTNs. From 2014 to 2019, a sample size of 1289 patients treated by RFA were 226 ones with solid nodules and 865 ones with cystic-solid nodule, respectively. The ecacy including the nodule maximal diameter reduction ratio (MDRR), the volume reduction ratio (VRR) and the cosmetic scores reduction ratio (CSRR). The results of the nodule MDRR and VRR in the cystic-solid nodule group were signicantly better than those in the solid nodule group at the 3rd and 6th month, and the CSRR in the two groups showed statistically signicant difference at the 3rd month. In a word, RFA is an effective method for symptomatic benign solid or cystic-solid nodules. The achieved MDRR and VRR in the cystic-solid nodule group were signicantly better than those in the solid nodule group at the 3rd and 6th month.


Introduction
Thyroid nodules are a common nding in general population. They can be approximatively discovered in 20%-76% by ultrasound (US) scan, and in 3%-7% by palpation 1 . Most nodules are asymptomatic and benign, and can be managed by observation only. However, even benign nodules may cause problems, such as symptoms of tracheal and esophageal compression, secondary to or combine with hyperthyroidism, retrosternal goiter, cosmetic concerns and psychological burdens 2 . The e cacy of drug therapy for thyroid nodules is not obvious, surgery and radioactive iodine therapy are still considered the mainstay of treatment, but both of these options have drawbacks. Radioiodine therapy is mainly aimed at the benign nodules of thyroid adenoma with hyperthyroidism, however, it is easy to be associated with hypothyroidism. In particular, surgery carries a 2%-10% complication rate and requires general anesthesia and hospitalization 3 . Furthermore, it is costly, and may cause problems such as scar formation and iatrogenic hypothyroidism. Ultrasound-guided minimally invasive treatment has the advantages of less trauma, less complications, no scar on the neck, short treatment time, safe, effective, repeatable treatment etc., and is generally used in clinic. Minimally invasive therapy is mainly divided into chemical ablation and thermal ablation. Thermal ablation refers to radiofrequency ablation (RFA), microwave ablation (MWA), laser ablation (LA) and high-intensity focused ultrasound (HIFU) [4][5][6][7] .
Ultrasound-guided RFA has exhibited distinct e cacy for benign thyroid nodules via decreasing the nodular volume by 84.8% 3,8 . At present, the indications of radiofrequency ablation are gradually increasing, including parathyroid hyperplasia, papillary thyroid microcarcinoma and metastatic lymph nodes. Thermal ablation is considered a better alternative therapy for single papillary thyroid carcinoma without capsule invasion or lateral cervical metastasis. Although surgical resection is the rst choice for patients with recurrent thyroid cancer, if the patients have severe scars and broplasia in the operation area and cannot tolerate or refuse second surgery, radiofrequency ablation of the cancer focus and related lymph nodes can be performed. According to a recent meta-analysis of RFA for locally recurrent thyroid cancer, the curative success is 100%, with a serum thyroglobulin decrease of 71.6% 9 . This retrospective observational study compared the clinical outcomes after a single RFA of solid nodules vs. with a high-frequency linear probe (L9-3) was used to monitor the RFA procedures ( Fig. 2A, 2D), as well as the initial diagnostic evaluation and the follow-up 11 . All patients were studied with US before the treatment by one of two radiologists (Kefei Cui and Yuanjing Huang) enrolled in the study, with 35 and 20 years of thyroid US experience, respectively.
Pre-ablation assessment. Examination of the bilateral vocal cords with a laryngo-berscope and blood coagulation tests were performed before RFA and the values were within normal range in all patients. The cosmetic score 10,12 was evaluated by an experienced clinician as follows: (1) no palpable mass; (2) no cosmetic problem but palpable mass; (3) a cosmetic problem on swallowing only; (4) easily visible mass.
Image and clinical analysis. All medical records and US images were reviewed by experienced radiologists. Cosmetic and symptomatic problems of the patients and location, largest diameter, volume and the proportion of the solid component of each nodule were evaluated before and after RFA. For each thyroid nodule, the pre-and post-treatment volumes 10,12 were calculated as V = πabc/6 (where V is the volume, a is the largest diameter, b and c are the two other perpendicular diameter).
RFA Procedure. The RFA system of the VIVA RF generator (VIVA RF generator, STARmed, Goyang, Korea) were applied in the study. An 18-guage, monopolar, modi ed, internal-cooled RFA antenna (VIVA, STARmed, Goyang, Korea) with a 1-cm active tip and a 7-cm shaft length was applied, which was speci cally modi ed for the ablation of thyroid nodules 12 .
Patients were put in a supine position with hyperextended neck. Two ground pads were positioned on the patient's thighs before the procedure. We used trans-isthmic and moving-shot techniques 15 . All procedures were completed under a sterile operation and local anesthesia with 2% lidocaine. An electrode was inserted in the same position of the local anesthesia site under US guidance, with the electrode tip initially positioned in the deepest and remote region of the lesion (Fig. 2B). The "moving shot technique" consists in the introduction of an internally cooled needle with variable diameter (17-19G), length (7-15cm) and active tips (ranging from 0.5mm to 2cm). The needle is inserted through the isthmus (transisthmic approach), starting from the middle to the lateral direction, to reach the nodule which is divided in different small hypothetical zones, each ablated by moving the tip of the needle from the deepest position upwards to the most super cial part of the nodule. Finally, the ablation of the lesion is con rmed by the US appearance of a hyperechogenic area associated to a sudden increase of impedence registered by the external generator (Fig. 2C).
For mixed or mainly cystic nodules, uid was aspirated under US-guided with a 18 gauge needle just before ablation. When the nodule was located in the upper or lower pole of the thyroid or adjacent to vital structures such as the vagus nerve, trachea or esophagus, hydrodissection 24 , consisting of 5% dextrose solution injection between the peripheral nodule area and surrounding critical structure, is aimed at avoiding thermal injury of adjacent structures (Fig. 3).
Statistical analysis. Data analysis was performed using the SPSS software (SPSS for windows 21.0, SPSS, Chicago, IL) and GraphPad Prism version 5.0 (California, USA). Continuous variables were as the mean ± standard deviation (SD). One-way ANOVA and the Student's t test were used to analyze MDRR, VRR and CSRR. The c2 test was applied for categorical variables. A value of p < 0.05 was considered to de ne statistically signi cant difference.

Results
From May 2014 to September 2019, 1289 patients (198 men, 1091 women) with BTNs were admitted in the First A liated Hospital of Zhengzhou University according to the inclusion criteria and underwent RFA procedures. All patients' thyroid functions including TSH, FT3 and FT4 were at normal ranges before ablation. Patient demographics and clinical features of appear in Table 1. For solid group patients, the mean maximal diameter, volume and cosmetic scores of nodules before RFA were 2.88 ± 1.39cm, 5.60 ± 4.52ml, 2.35 ± 0.97. For cystic-solid group patients, the mean maximal diameter, volume and cosmetic scores of nodules before RFA were 3.19 ± 1.47cm, 8.12 ± 6.32ml, 2.58 ± 0.94. Comparison of these three variables was different among the two groups (p<0.05). In addition, there were no signi cant differences in the gender, age and follow-up time (p>0.05). Based on the differences among base-lines, we only chose MDRR, VRR and CSRR to analyze the two groups, while for solid nodules or cystic-solid nodules themselves, we supplemented the analysis with data and reduction rate differences at the 3rd, 6th, 12th month. In the solid group, the mean maximal diameters of nodules at the 3rd, 6th, 12th month were 2.12 ± 0.89cm, 2.04 ± 0.76cm, 1.46 ± 0.72cm, respectively, all of which were signi cantly less than that before RFA (p < 0.001) (Fig. 4A); the volume of nodules at the 3rd, 6th, 12th month were 3.01 ± 2.75ml, 2.40 ± 2.19ml, 0.59 ± 0.58ml, respectively, all of which were signi cantly less than that before RFA (p < 0.001) (Fig. 4B); the cosmetic scores of nodules at the 3rd, 6th, 12th month were 1.68 ± 0.80, 1.61 ± 0.73, 1.25 ± 0.45, respectively, all of which were signi cantly less than that before RFA (p < 0.001) (Fig. 4C). Through further pairwise comparison, we found that there were no signi cant difference between 6mo vs. 3mo and 12mo vs. 6 mo after RFA (p > 0.05). While 12mo vs. 3mo showed signi cant difference after RFA (p < 0.05), indicating that the solid nodule group shrank markedly at 12mo after operation.
To further verify above results, we selected MDRR, VRR and CSRR associated with nodular maximal diameter (MD), volume (V) and cosmetic score (CS) as study subjects. The same differences could be observed in MDRR (Fig. 4D) and VRR (Fig. 4E). However, CSRR changed signi cantly at 6mo, which may be explained by the variation of interior and exterior. The maximum diameter and volume of solid nodules have been shrinking after ablation, both of which can be calculated. However, the cosmetic score is an external performance, even though the nodules are still shrinking at 12mo, the appearance has been improved greatly at 6mo in advance. Consequently, we could nd that there were no signi cant differences between 3mo vs. 12mo and 6mo vs. 12mo after RFA (p > 0.05). Nevertheless, 6mo vs. 3mo showed signi cant difference after RFA (p < 0.05), the cosmetic scores were almost consistent at 6mo and 12mo (Fig. 4F). Taken together, these results demonstrated that the maximum diameter and volume of solid nodules group would become smaller and smaller with the extension of time, both shrank prominently at 12mo during the observation period.
Some studies have found that larger MDRR and VRR can be achieved in the RFA group at 6 months 13 . Interestingly, similar results were observed in cystic-solid nodules group. By further analysis, we found that the rangeability of MDRR (Fig. 5D) and VRR (Fig. 5E) was consistent to CSRR (Fig. 5F), all decreased obviously at 6mo. The reason why these three variables showed synchronously is due to the different ablation method of cystic-solid nodules. The cystic uid was aspirated before ablation, which may arti cially accelerate the process of nodular shrinkage. Hence, the decline peak of nodular maximum diameter and volume appears in advance, synchronized with the cosmetic score. These results suggest that cystic-solid nodules shrank obviously at 6mo after RFA.
In order to further compare the e ciency of the two groups after ablation, MDRR, VRR and CSRR were also selected as study subjects.  (Fig. 6C).
The mean nodular MDRRs and VRRs at the 3rd, 6th months showed signi cant difference between the two groups, but at the 12th month, statistical signi cance was not found, and the cystic-solid group had a higher MDRR and VRR than that in the solid group. Between the two groups, there was no signi cant difference in the mean CSRR at the 6th, 12th month. While at the 3rd month of the follow-up, statistical signi cance was found, and the cystic-solid group had a higher CSRR than that in the solid group.
Those results above may be attributed to different ablation method and transformation of nodular nature. The aspiration of cystic uid could arti cially increase the reduction ratio of cystic-solid nodules, which was higher than that of the solid nodules. Six months after ablation, cystic-solid nodules can completely convert into solid nodules, giving rise to no signi cant difference at 12mo. Therefore, the 6mo can be regarded as the con uent point of cystic-solid and solid nodules. Previously, the decrease rate of the two groups was different from each other, owing to their diverse composition. While after 6 months, the two groups performed synchronous rate of change due to the shift from cystic-solid to solid, and we can see that the difference was not signi cant at 12mo.
The whole set of data was applied to testify that BTNs decreased most signi cantly at the 6th month after RFA 13 . Six comparative indices (MD, V, C, MDRR, VRR and CSRR) were selected for analysis. In the whole benign thyroid nodules group, the mean maximal diameter, volume and cosmetic scores of nodules before RFA were 3  7E). All the above comparative indices before ablation vs. the observation time in the BTNs group signi cantly decreased; the P values were less than 0.001.
By pairwise comparison analysis, we found that there was statistical signi cance between 6mo vs. 3mo after RFA (p < 0.05), while 12mo vs. 6mo had no signi cant difference after RFA (P > 0.05), indicating that the BTNs shrank obviously at 6mo after operation. Comparing solid nodules, cystic-solid nodules and all benign thyroid nodules, we found that during the one-year follow-up period, solid nodules shrank most signi cantly at 12mo. Whereas, cystic-solid nodules and benign thyroid nodules shrank most signi cantly at 6mo, which is in line with most studies' results 13 .

Discussion
Our study demonstrated the e cacy of RFA according to the types of benign thyroid nodular with a oneyear observation time. In this study, RFA effectively reduced nodular size by 90% approximately. It is showed that RFA can be an effective way for the treatment of benign thyroid nodules 14 .
For MDRR and VRR, no signi cant differences were found after 12 months between the two groups, while the cystic-solid group achieved better results than the solid group at the 3rd and 6th month. This could be explained by the aspiration of cystic uid and the transformation of nodular characteristic. Due to the aspiration of cystic uid, the MDRR and VRR of cystic-solid nodules exhibited greater increase than those of the solid nodules. In addition, in light of the shift from cystic-solid nodules to solid nodules after ablation, the result showed that there was no signi cant difference between the two groups at the 12th month.
In our research, there is an inconsistency between CSRR and MDRR or VRR. For instance, the maximum diameter and volume of solid nodules decreased most signi cantly at the 12th month, while the cosmetic scores improved most signi cantly at the 6th month. We interpret this phenomenon as the difference between calculation and sensation. CSRR improved obviously at the 6th month in terms of appearance, but it does not represent the peak of nodule maximum diameter and volume reduction ratio (only judged by calculation). Of course, there are also consistent situation, such as cystic-solid nodules, the cosmetic improvement accords with the peak of the maximum diameter or volume reduction rate. This may also be ascribed to the arti cial acceleration of size change after aspiration of cystic uid, thus moving the VRR' peak forward, in line with the CSRR. Throughout the full data, the achieved conclusion between benign thyroid nodules and cystic-solid nodules was similar. It may be related to the fact that most of the benign nodules found in our daily life are cystic-solid, while malignant nodules are usually solid and not suitable for ablation.
During the RFA procedure, we adopted the "moving-shot technique". It is more exible than " xed-needle technique", and the size of the ablation unit can be adjusted by changing the rate of tip movement 15 . The vital structures around the nodule can be protected by accelerating the movement speed and using a relatively small ablation unit, while the safe area in the center of the nodule can prolong the action time and adopt a larger ablation unit. For the position with rich blood ow, tissue necrosis can be made more thoroughly by increasing the power and slowing down the movement speed or even transitioning to xed ablation, in order to ensure the safety and long-term effect of ablation. On the contrary, single point xedneedle technique can only extend the ablation area by prolonging the action time, and the shape of ablation unit can not be changed. As a result, the ablation is often incomplete. Incomplete ablation usually leads to inadequate shrinkage of nodules and pathological hyperplasia in the margin of the residue, which may be related to thermal ablation stimulation. The method of multi-point and multineedle not only increases the risk of injecting more times, but also hardly achieve complete ablation, because there are many vital structures (laryngeal recurrent nerve, trachea, esophagus) around the thyroid gland 16 . Moreover, It is dangerous for the critical tissue to prolong the action time of the xed needle tip.
To date, RFA has gained widespread use as an alternative to surgery for solid malignancies including the liver, kidneys and lungs [17][18][19] . The rst experience of image-guided RFA with evaluation of e cacy and safety of ultrasound guided percutaneous RFA of thyroid nodules has been reported and applied in 2006 20 . During the procedure, a RF generator is used to produce voltage between an active electrode (applicator) and a reference electrode (grounding pad). The voltage is used to establish an oscillating electric eld, which in turn induces frictional heating by causing electrons to collide with nearby molecules nearest to the applicator. Tissue heating to temperatures greater than 60 ˚C causes immediate coagulation necrosis. Nevertheless, it could lead to evaporation and carbide and make damages to the surrounding tissue if the temperature is higher than 100 ˚C.
RFA-associated complications have been reported, including voice change, skin burns, hematoma formation, vomiting, brachial plexus injury, nodule rupture, permanent hypothyroidism and transient hyperthyroidism 21,22 . Reported the incidence of voice change after RFA is higher for recurrent thyroid cancers (7.95%) than for benign thyroid nodules (0.94%), possibly because of the absence of a safety area around recurrent tumors 23 . Voice changes are likely due to recurrent laryngeal nerve dysfunction, which can be caused by thermal injury. Thermal injury to the recurrent laryngeal nerve may be prevented by using the hydrodissection with continuous uid infusion 24 . Before RFA, the nodules adjacent to the "dangerous triangle" of the neck or the medial capsule of the thyroid gland should be injected with local anesthetic or normal saline to form a liquid isolation zone, which can effectively prevent the thermal injury to the nerve during the process. Part of transient hoarseness dates back from the blocking effect of local anesthetic on recurrent laryngeal nerve, which can be recovered within a few hours. For the super cial nodules close to the thyroid capsule, the heat can easily spread to the anterior cervical muscles and cause neck pain during ablation. This phenomenon can also be avoided by hydrodissection 24 . Although the incidence of complications of radiofrequency ablation is low, it should be paid attention to. Complications can be prevented or properly handled by deepening understanding and mastering skills.
Voice changes after EA (ethanol ablation, EA) are very rare. Although the exact incidence of voice changes after EA is unclear, it is lower than the incidence of voice changes after RFA. A possible mechanism for voice changes is the leakage of ethanol outside the thyroid gland, causing damage to the recurrent laryngeal nerve. As compared with RFA, EA has the bene ts of reduced cost, pain and risk of developing nerve injury. However, results of the previous study comparing the e cacy of the two treatment modalities for hepatocellular carcinoma have suggested that RFA may be more effective in reducing the volume, because it can produce larger zone of tumor destruction than EA 25,26 . Local tumor recurrence rates of 0%-25% and 3.2%-33% have been reported after RFA and EA for recurrent thyroid cancers, respectively 27 .
Both RFA and MWA were certi ed effective and safe in treating BTNs 28 . Liu et al. 29  Several studies compared LA to RFA and showed that the nodular VRR for RFA is higher than that for LA 32,33 . While in favor of LA is a recent study performed by Pacella et al. 34 where in 601 nodules, a signi cant nodule shrinkage was observed in the LA group, particularly in large nodules. No matter which method has more advantages, RFA and LA are all safe and effective measures for their low complication rates, high nodular VRRs and the preservation of thyroid function [35][36][37][38][39] . Zhou et al. 40 conducted a retrospective study on solitary papillary thyroid carcinoma (36 received LA and 45 underwent surgery). No signi cant difference was found in the recurrence rates between the LA and surgical group (5.6% vs. 6.7%). LA can be considered as an alternative in the local control of papillary thyroid microcarcinoma 41 .
HIFU is a new minimally invasive technique, which was rst applied to clinic in 2010. The coagulative necrosis in targeted zone occurs through a focused high energy ultrasound beam. Recently, HIFU has been used to treat breast cancer and uterine leiomyoma. However, this technique is still in its infancy for the therapy of benign or malignant thyroid nodules. Compared with RFA and MWA, HIFU has lower incidence of infection and higher acceptance of patients, which also requires less anesthetic dose.
Kovatcheva et al. 42 evaluated the e cacy and safety of HIFU in the treatment of benign solid nodules, the results showed that the average volume decreased by 48.7% ± 24.3% after 6 months, and only 2 patients presented mild subcutaneous edema and skin redness.
Compared with EA, MWA and LA, RFA is more widely used in clinic for its superiority in volume reduction 43,44 . Ultrasound-guided radiofrequency ablation of benign thyroid nodules is a technical innovation, which is safer and more minimally invasive compared with surgical resection. Some researchers pointed out that the operator's experience is the key to therapeutic effect 43 . But beyond that, we should strictly master the indications 45 , eliminate contraindications, perfect the preoperative examination and standardize the operation. At present time, RFA is recommended for the treatment of low-risk papillary thyroid microcarcinoma because so far there is some scienti c evidences on its effectiveness and safety [46][47][48] .
Our study has several limitations. First, the vascular score was not evaluated and analyzed before and after RFA. Second, the main one was the short follow-up. Moreover, this study was a retrospective design and its performance at a single center. A future prospective, longer follow-up study is necessary to con rm our preliminary results.
In conclusion, RFA is an effective technique for patients with BTNs due to its signi cant shrinkage of nodular volumes and the apparent improvement of cosmetic concerns related to the nodules. The achieved MDRR and VRR in the cystic-solid nodule group were signi cantly better than those in the solid nodule group at the 3rd and 6th month. The hydrodissection with continuous uid infusion. A buffering zone (green dotted line) was created to establish a liquid isolation zone > 0.5cm in depth between the thyroid and adjacent structures.