Comparison of endoscopic bilateral areolar and robotic-assisted bilateral axillo-breast approach thyroidectomy in differentiated thyroid carcinoma: a propensity-matched retrospective cohort study

We retrospectively reviewed 1357 patients who underwent thyroidectomy using endoscopic or robotic-assisted procedures from November 2018 to September 2021 in the 960th PLA Hospital (Ji Nan, China). A total of 1030 patients were excluded based on the following criteria: Benign (n=221), lateral neck dissection (n=655), Non- DTC carcinoma (n=7), Lateral neck node metastases (n=2), Distant metastasis and invasion to adjacent organs of the thyroid (n=1), Previous neck or thyroid surgery(n=10) and Non-anterior chest approach (n=86) or transoral approach (n=48), we were recommended to evaluate the thyroid gland as well as central and lateral neck lymph node chains using the preoperative neck ultrasonography as the initial step [15]. Then, preoperative laryngoscopy was used for routine assessment of the vocal fold motion, and thyroid function tests were performed to assess thyroid function abnormalities [16, 17]. Following the ATA and NCCN guidelines, the initial surgical procedure should be a thyroid lobectomy unless there are clear indications to remove the contralateral lobe [3, 18, 19]. Therefore, this study included patients who underwent lobectomy or total thyroidectomy with central neck dissection (CND). Through adequate preoperative explanation of BAA in ET and BABA in RT, the operation method was decided according to the patient’s preference. Patients were categorized into two groups according to the type of surgery. The medical charts and pathology reports of 327 patients were reviewed and analyzed using retrospective cohort study methods. After propensity score matching(PSM), we eventually included 177 patients who underwent thyroidectomy with CND for DTC in this study. Forty-two patients were included in the ET group, and 135 patients were included in the RT group. A flow chart of the study is presented in Fig. 1. Ethical approval for this study (Ethical Committee NO.157) was provided by the Ethical Committee of the 960th Hospital of the People’s Liberation Army, Jinan, China on 26 December 2022. Due to the retrospective nature of this study, the Ethical Committee waived the requirement for informed consent.

In our institutions, three surgeons with > 10 years of experience in thyroid and endocrine surgery were involved in the operation. All thyroidectomies were performed by the same experienced surgeon teams with the da Vinci Si Surgical Robot System (Intuitive Surgical, Inc., Sunnyvale, CA). A neural integrity monitor was utilized for all the surgery [20]. The day before the operation, 0.05-0.1 ml of carbon nanoparticles’ suspension injection was injected into the left and right lobe thyroids under the guidance of ultrasound to leave the parathyroid glands unstained and staining level VI area lymph nodes. Previous studies have demonstrated that robotic and endoscopic procedures are very similar, except for using a robotic arm [12, 13]. The difference is that the BABA approach has more incisions in the axilla than the BAA approach [9]. Therefore, our study’s endoscopic and robotic thyroidectomy surgical procedures were virtually identical. Therapeutic CND when suspicious lymph node enlargement is detected on preoperative or intraoperative examination, prophylactic CND is for patients with clinically uninvolved central neck PTC, especially for advanced primary tumors [18].

The BABA approach was performed in RG. A 1.5 cm incision was made at the internal edge of the right areola for the insertion of the scope of the surgical system, an incision was made at the edge of the left areola and anterior of the right axilla for the placement of an 8 mm trocar as operating approach, an incision was made anterior of the left axilla for the post of a 5 mm trocar as operational approach, and an incision was made anterior of the right areola for the insertion of a 12 mm trocar as a scope. A robotic surgical system was connected to establish the operating space, and the working area was maintained by CO₂ with a pressure of 5-6 mmHg. The isthmus of the thyroid was cut with a harmonic scalpel, and the side of the lesion was resected. The bilateral recurrent laryngeal nerves (RLN) were protected using intraoperative neuromonitoring (IONM) [20].

The BAA approach was performed in EG. A 12 mm trocar was inserted into the right areola incision to access the endoscope lens, and 8 mm trocars were inserted into the left and right areola incisions to place the operating instruments. The operational space was maintained by inflation with CO2 at a pressure of 5 to 6 mmHg. The harmonic scalpel was used to divide the isthmus of the thyroid. The thyroid on the focal side was resected. At the same time, the RLN was monitored using the IONM, to preserve the parathyroid glands and protect the RLN. The details of the BABA robotic procedure and the BAA endoscopic procedure were described in the previous report [9, 10].

For comparison of pathological characteristics and surgical outcomes between the two groups, we measured the maximum diameter of the thyroid, number of positive lymph nodes metastasis, total operation time, the total amount of drainage, drain insertion, postoperative pain score (VAS), cosmetic satisfaction, postoperative hospitalization stay, number of retrieved lymph nodes, and postoperative complications. Postoperative laboratory data such as serum calcium and parathyroid hormone were also measured and compared. Hypoparathyroidism was defined as a serum parathyroid hormone (PTH) <15.0 pg/ml, and hypocalcemia was defined as serum calcium <2.08 ng/mL or having positive symptoms after thyroidectomy. The size of the thyroid specimen was determined by the longest diameter of the most oversized thyroid in cases of multiple specimens. The operative time was determined from skin incision to closure. The postoperative pain was measured using the visual analog scale (VAS) score (0–10 points) on the postoperative days; The period of hospitalization was defined from the day of surgery to the day of discharge. RLN injury was defined as vocal cord palsy persisting after six months. Cosmetic satisfaction was assessed on a scale from 0 (dissatisfied) to 3 (extremely satisfied).

To minimize the effects of selection bias, propensity score matching was performed using nine clinicopathological characteristics: age, gender, BMI, tumor size, extrathyroidal extension (ETE), multifocality, the extent of surgery, type of carcinoma, and combined thyroiditis. Individual patient propensity scores were calculated using logistic regression analysis (Fig. 2). Forty-two patients who underwent ET were matched in a 1:4 ratio with 135 patients who underwent RT. Although we presented the TNM stage in Table 1, we did not use the TNM stage as a matching variable for the following reasons. First, the components of the T-stage (tumor size and ETE) were used as matching variables. Then, we initially excluded DTC patients with distant metastasis, including lateral neck node metastases. Finally, retrieved CLN was used as an outcome variable in this study to compare the efficacy of the two groups.

The baseline clinical and pathological characteristics of 327 patients before PSM are shown in Table 1. The patients were divided into two groups, including 278 robotic procedures (RG) and 49 endoscopic procedures (EG). There were more male patients in the RG (p = 0.002). The patients who performed also had a higher ratio of T1a category, a lower ratio of T1b category, and T3b category in RG (P= 0.003). Furthermore, the pathological types differed significantly in the two groups (P=0.048). Although the EG had a larger tumor size and more extrathyroidal extension cases than the RG, there were no significant differences found in the two groups (6.55±5.25 mm in EG vs. 5.15±2.67 mm in RG, p = 0.331 and 12.2% in EG vs. 5% in RG p = 0.052, respectively), The other variables such as age (p = 0.561), BMI (p = 0.586), multiplicity (p = 0.832), tumor location (p = 0.124) did not show a statistical difference. The results suggest that the baseline levels of the two groups of data are inconsistent. By using 1:4 propensity score matching, we minimized the bias in baseline data between the two groups. Table 2 shows the two groups’ baseline clinical and pathological characteristics after propensity matching. The EG had 42 patients, and 135 patients underwent RT. There was no significant difference in baseline data between the two groups after propensity matching.

Table 3 shows the surgical outcomes of 177 patients after propensity score matching. The mean operation time was significantly longer in the EG than in the RG (207.5 ± 50.8 min in EG vs. 132.9 ± 43.1 min in RG, p < 0.001). Furthermore, more retrieved central lymph nodes were removed by RG (5.4 ± 3.6 in EG vs. 11.9 ± 6.1 in RG, p < 0.001). Besides, the mean maximum diameter of the thyroid was more expansive in the EG than in the RG (5.15 ± 0.90 cm in EG vs. 4.93 ± 0.91 cm in RG, p = 0.04). There were no significant differences in the total amount of drainage and drain insertion between the two groups (239.8 ± 100.0 ml in EG vs. 266.0 ± 150.2 ml in RG, p = 0.241 and 5.79 ± 1.6 day in EG vs. 6.15 ± 2.04 day in RG p = 0.316, respectively). Both groups showed that cosmetic satisfaction (p = 0.837) and postoperative hospitalized stay (p = 0.100) were similar. Furthermore, positive LN metastasis (p =0.561) and pain score (p = 0.077) showed no significant difference between the two groups.

The relationship between the operation time and the number of operations is shown in Fig. 3. One hundred thirty-five patients underwent robotic-assisted thyroidectomy in this study. Using the simple linear regression model and smooth curve (moving average time), the operation times for RG trended to decrease gradually as the cases went by. (95% CI -0.4735 to -0.1090, P= 0.002).

Table 4 shows the two groups’ postoperative complications and postoperative laboratory data. The mean postoperative PTH and total calcium were nonsignificantly between the 2 groups (21.1± 12.4 pg/mL in EG vs 20.7 ± 15.1 pg/mL in RG P=0.659, 2.19 ± 0.14 ng/mL in EG vs2.14 ± 0.13 ng/mL in RG P=0.068, respectively). The two groups had the same rate of hypoparathyroidism during the surgery (p = 0.331). There was less hypocalcemia in the EG, but there was no significant difference between the two groups (19 % in ET vs. 34.1 % in RT, p = 0.065). Only one patient in the RG developed seroma (0.7%). Besides, two patients developed chyle leakage (1.5%), but both groups observed no significant infection, bleeding, RLN injury or flap necrosis during hospitalization.