Safety and Efficacy of Transoral Robotic Thyroidectomy for Thyroid Tumor: A Systematic Review and Meta-Analysis

Simple Summary This systemic review and meta-analysis compared and analyzed the safety and effectiveness of transoral robotic thyroidectomy on the thyroid tumor with other thyroid approaches. Transoral robotic thyroidectomy showed similar results to other robotic-assisted thyroid surgeries. Compared to a conventional open thyroidectomy, transoral robotic thyroidectomy had longer operational times and hospitalization days, and worse postoperative pain, but a higher cosmetic satiation score. However, more randomized controlled studies need to be included and analyzed. Abstract Background: To assess the safety and effectiveness of transoral robotic thyroidectomy (TORT) in thyroid tumor. Methods: PubMed, Embase, Web of Science, SCOPUS, Cochrane database, and Google Scholar up to June 2022. Studies comparing outcomes and complications between TORT and control groups (robotic bilateral axillo-breast, trans-axillary, postauricular approach, conventional open thyroidectomy (OT), and transoral endoscopic approach) were analyzed. Results: Ten studies of 1420 individuals. The operative time (SMD 1.15, 95%CI [0.48; 1.89]) was significantly longer and the number of retrieved lymph nodes (LNs) (SMD −0.27, 95%CI [−0.39; −0.16]) was fewer in TORT than in the control group. The postoperative cosmetic satisfaction score (SMD 0.60, 95%CI [0.28; 0.92]) was statistically higher in TORT than in the control group. In subgroup analysis, there was no significant difference between robotic surgeries. However, TORT had significantly longer operative times (SMD 2.08, 95%CI [0.95; 3.20]) and fewer retrieved LNs (SMD −0.32, 95%CI [−0.46; −0.17]) than OT. TORT satisfied significantly more patients in cosmetic view. However, it increased hospitalization days and postoperative pain on the operation day and first day compared to OT. Conclusions: TORT is not inferior to other robotic-assisted approaches. Its operation time and hospitalization days are longer and postoperative pain is greater than OT, although its cosmetic satisfaction is high.


Introduction
The prevalence of thyroid tumors, including thyroid cancer, has been increasing for several decades [1]. With the development of diagnostic tools for small thyroid lesions, thyroid surgery from lobectomy to total thyroidectomy has increased [2,3]. In the past, the standard treatment for thyroid tumors was conventional open thyroidectomy (OT) [4]. Although conventional OT exposes the thyroid gland sufficiently, it can leave a large scar on the neck [5]. Most patients with thyroid nodules are women. About 5% of females have thyroid nodules [1]. Most patients with thyroid tumors, especially young female patients, are concerned about neck scars [6,7]. Various endoscopic and robotic thyroid surgeries have been studied for decades to avoid anterior neck incision [4,5].
Thus, the objective of this study was to perform a meta-analysis to evaluate the safety and effectiveness of TORT for thyroid tumor compared to traditional OT, robotic BABA, and TOETVA for operative outcomes, postoperative outcomes, and complications. To the best of our knowledge, this is the first meta-analysis that directly compares TORT with other approaches for thyroid tumors.

Search Strategy
PubMed, SCOPUS, Embase, the Web of Science, Google Scholar, and the Cochrane database were searched to retrieve prospective or retrospective articles published in English prior to June 2022. Thyroidectomy, transoral thyroidectomy, remote-access thyroid surgery, minimally invasive surgery, robotic thyroidectomy, robotic thyroid surgery, surgical approaches, transoral approach, vestibular approach, transoral-vestibular robotic thyroidectomy, bilateral axillo-breast approach, open thyroidectomy, thyroid neoplasm, thyroid carcinoma, thyroid nodule, cosmesis, and comparison were used as search terms and keywords. Two independent literature reviewers reviewed and screened titles and abstracts of all searched studies, excluding studies about patients with thyroid neoplasm, not related to TORT. If the abstract alone could not determine whether or not the study should be included, the full text was then checked to determine its eligibility. Figure 1 presents a flow chart for selecting eligible studies to be analyzed. We registered the study protocol on Open Science Framework (https://osf.io/gtnck/) (accessed on 25 July 2022).

Data Extraction and Risk of Bias Assessment
Data were extracted from selected eligible studies regarding the number of patients, scale and score used for cosmetic satisfaction, operative time, hospitalization day, postoperative pain score, retrieved lymph node (LN) number, incidence or percentage of postoperative bleeding, skin flap perforation, postoperative hypoparathyroidism, vocal cord palsy, incidental parathyroidectomy, seroma, and p-value in comparison between treatment (TORT) group and control group (other types of thyroidectomy, including robotic BABA, robotic PAA, conventional OT, TOETVA). Data were then organized using a standardized format [17][18][19]21,[24][25][26][27][28][29][30]. The Newcastle-Ottawa Scale was used to evaluate non-randomized controlled studies [31,32].

Statistical Analysis
Meta-analyses were conducted using 'R' statistical software (R Foundation for Statistical Computing, Vienna, Austria). When representing the original data as a continuous variable, meta-analysis was performed using standard mean difference (SMD). For all other cases, incidence analyses were performed using odds ratio (OR). Sensitivity analyses were also conducted to assess the effect of each study on overall meta-analysis results. Cancers 2022, 14, x 3 of 17

Data Extraction and Risk of Bias Assessment
Data were extracted from selected eligible studies regarding the number of patients, scale and score used for cosmetic satisfaction, operative time, hospitalization day, postoperative pain score, retrieved lymph node (LN) number, incidence or percentage of postoperative bleeding, skin flap perforation, postoperative hypoparathyroidism, vocal cord palsy, incidental parathyroidectomy, seroma, and p-value in comparison between treatment (TORT) group and control group (other types of thyroidectomy, including robotic BABA, robotic PAA, conventional OT, TOETVA). Data were then organized using a standardized format [17][18][19]21,[24][25][26][27][28][29][30]. The Newcastle-Ottawa Scale was used to evaluate non-randomized controlled studies [31,32].

Statistical Analysis
Meta-analyses were conducted using 'R' statistical software (R Foundation for Statistical Computing, Vienna, Austria). When representing the original data as a continuous variable, meta-analysis was performed using standard mean difference (SMD). For all other cases, incidence analyses were performed using odds ratio (OR). Sensitivity analyses were also conducted to assess the effect of each study on overall meta-analysis results.

Results
Ten studies of 1420 individuals were included. Study characteristics are summarized in Table 1. Egger's test and Begg's funnel plot analyses revealed no publication bias for operative time (OR 0.4171), retrieved LN number (OR 0.6468), or incidence of temporary

Results
Ten studies of 1420 individuals were included. Study characteristics are summarized in Table 1. Egger's test and Begg's funnel plot analyses revealed no publication bias for operative time (OR 0.4171), retrieved LN number (OR 0.6468), or incidence of temporary vocal cord palsy (OR 0.19) in the included studies ( Figure 2). In contrast, the number of included studies for cosmetic satisfaction, hospital day, pain score, and incidence of bleeding, flap perforation, parathyroidectomy, seroma, permanent hypoparathyroidism, and vocal cord palsy was insufficient to adequately obtain a funnel plot or perform advanced regression-based assessments. Thus, publication bias was not evaluated for them.

Operation-Related Outcomes between Treatment Group and Control Group
The operative time (SMD

Operation-Related Outcomes between Treatment Group and Control Group
The   There was a significant heterogeneity (I 2 > 50%) in operation-related measurements because control groups involved different types of approaches. Therefore, subgroup analysis was conducted to assess the change in comparative advantage or disadvantage of TORT according to the type of operation. In subgroup analysis, there were no significant differences in some operation-related measurements between robotic surgeries (TORT, transaxillary robotic approach, and robotic PAA). However, the TORT group had significantly longer operative times and fewer LNs retrieved than the conventional OT group (SMD 2.

Peri-Operative Complications between the Treatment Group and Control Group
Incidences of postoperative bleeding (hematoma) ( Figure 4). However, the postoperative infection rate (OR 10.67, 95% CI [1.24; 91.66]; I 2 = 0.4%) was significantly higher in the TORT group than in the control group. There was no significant heterogeneity (I 2 < 50%) in postoperative complications (Figure 4).

Postoperative Outcomes between the Treatment Group and Control Group
The postoperative cosmetic satisfaction score (SMD 0.60, 95% CI [0.28; 0.92]; I 2 = 75%) was statistically higher in the TORT group than in the control group. However, there was no significant difference in postoperative pain score on op day (SMD −1.14, 95%  There was a significant heterogeneity (I 2 > 50%) in operation-related measurements because control groups involved different types of approaches. Therefore, subgroup analysis was conducted to assess the change in comparative advantage or disadvantage of TORT according to the type of operation. In subgroup analysis, there were no significant differences in some operation-related measurements between robotic surgeries (TORT, trans-axillary robotic approach, and robotic PAA). However, the TORT group had significantly longer operative times and fewer LNs retrieved than the conventional OT group (

Sensitivity Analyses
Sensitivity analyses were performed to determine differences in integrated estimates in such a way that the meta-analysis was repeated, excluding one study each time. All results were consistent with the results described above.
In our study, the operation time and retrieved LN number were significantly different between the TORT and the control group. The TORT group also showed significantly longer operation times and fewer retrieved LNs than the other approaches in the subgroup analysis. In general, in other surgeries as well as thyroid surgery, the surgical time of a robotic surgery is longer than that of a conventional OT [44,45]. This is because a roboticassisted surgery requires additional operation time to make flaps and dock the robotic arm and system to approach the thyroid tumor [29]. The fact that there were fewer retrieved LNs in the TORT group from our study might be because the clinician's accumulated experiences for TORT and conventional OT differed during the same period, resulting in a selection bias [46]. Because TORT is more newly introduced than conventional OT, the overall proficiency of TORT might be lower than that of OT. Robotic lateral neck dissection by a gasless unilateral axillo-breast approach is possible. The retrieved LN number can be changed if the clinician has more experience with the robotic-assisted approach and becomes skilled [47]. In addition, the reason why TORT has fewer retrieved LNs than conventional OT is probably that the insertion site in the oral cavity is narrow, or the carbon dioxide gas expansion makes the workspace unstable. This working environment can limit the motion range of the robotic arm for LN dissection. The type of LN dissection can also affect the retrieved LN numbers. Kim et al. have reported that the unilateral or bilateral central neck dissection type can affect retrieved LN numbers in thyroid cancer [46]. The robot-assisted approach does not have an incision at the lower central part of the neck. Thus, the central neck dissection range and retrieved LN number might be lower than those of OT [46]. For example, in robotic BABA, the central LN may not be sufficiently exposed because it is blocked by the clavicle [48]. In TORT, central neck dissection exposure could vary depending on the inflexible robotic arm and the camera. However, even though the retrieved LN number has been reported as a good prognostic factor in gastric or colon cancer surgery [49,50], whether the retrieved LN number affects survival and recurrence in thyroid cancer surgery is currently unclear [46].
According to our meta-analysis, the TORT group had a significantly higher postoperative cosmetic satisfaction score than the control group. Another study has also confirmed that cosmetic satisfaction with the robot-assisted approach is significantly higher than with conventional OT [34]. In particular, transoral thyroidectomy is a natural orifice transluminal endoscopic surgery technique that leaves no scar on the neck or body. In addition, it is less invasive than other minimally invasive thyroid surgery [51]. Although many extra-cervical approaches to the thyroid have been described to avoid anterior neck scarring, such as trans-axillary, robotic face lift, and so on, these incisions are not truly scar-free. Therefore, TORT tends to be more cosmetically effective than robotic BABA, almost reaching statistical significance (SMD 0.66, 95% CI [0.21; 1.10], p = 0.13).
However, this approach also has its own demerits beside the cosmetic aspect. Because the craniocaudal view and the workspace between both mental nerves are narrow, flap elevation might be technically challenging [51]. This difficulty causes this approach to exclude patients with a history of neck surgery or over-prominent mandible [30]. Because of this limited view and approach, patients with a highly placed upper pole of the thyroid, prominent Zuckerkandl tubercle, advanced thyroid cancer with tracheal or esophageal invasion, and posterior extrathyroid extension should be avoided. Wound changes from the external body surface to the oral cavity can also lead to numbness or paresthesia of the lower lip and chin skin (due to mental nerve injury) and postoperative infection (due to oral flora) [30]. Based on these, the clinician needs to decide which approach is suitable for patients accordingly.
From our results, the TORT group had significantly longer hospital stays than the OT group. The pain was also significantly more severe on the day of surgery and the first postoperative day in the TORT group than in the OT group. Compared to OT, TORT has a wider flap elevation with a working space from the intraoral incision and unilateral axilla to thyroid tumor [5], resulting in increased pain and a longer hospital stay. In another meta-analysis, the anterior chest pain score of a robot-assisted approach in the first week after surgery was also significantly higher than that of OT [52].
This meta-analysis has some limitations. First, since not many studies have directly analyzed TORT and other approaches for thyroid tumor, only a few studies were included in this meta-analysis. Second, most of the included studies were retrospective in nature. There were no randomized controlled trials. Thus, there might be a selection bias or reporting bias. Third, the incidence of complications was low in most groups, making it difficult to identify differences between TORT and control groups. Fourth, there might be heterogeneity of postoperative outcomes because postoperative management, such as surgical site compression, postoperative diet, and analgesic for pain control, differed depending on the institute. Lastly, postoperative pain was not evaluated by subgroup analysis depending on the location, such as the neck or anterior chest.

Conclusions
This meta-analysis found that TORT improved patients' cosmetic satisfaction but increased operative time, hospitalization days, and postoperative pain compared to conventional OT. However, since the number of studies included in this meta-analysis was small, conclusive confirmation of the clinical benefit of TORT needs to be studied further.

Conflicts of Interest:
The authors declare no conflict of interest.