1 INTRODUCTION

Thymectomy is used as a form of treatment for thymic lesions such as thymomas, thymic carcinomas, thymic cysts as well as myasthenia gravis (MG).¹ Traditionally, thymectomies have been performed by way of an open approach such as median sternotomy or thoracotomy with complete surgical resection (R0) as the standard of care.² Recently, minimally invasive techniques have also been developed for thymectomies including lateral video-assisted thoracoscopic surgery (LVATS), subxiphoid and robotic-assisted thymectomy.^2-4 Despite the uptake of minimally invasive surgery, there is an ongoing debate regarding the optimal mode of resection. Areas of concern are the potential for tumour seeding, lack of complete resection and locoregional recurrence when using a VATS approach.² However, LVATS thymectomy has been associated with lower post-operative mortality and morbidity, decreased post-operative complication rates, better cosmesis and shorter length of stay (LOS).^5-7

Within minimally invasive techniques, there is growing literature comparing the safety and effectiveness of LVATS with the emerging robotics and subxiphoid approach. In a meta-analysis that compared robotics with open (14 studies; n = 615 robotic and n = 2872 open) as well as robotics with VATS approach (seven studies; n = 428 robotic and n = 566 VATS), robotics was superior to open thymectomy in terms of having decreased blood loss, lower rates of post-operative complications and involved margins for resected thymomas, LOS and no differences in the duration of surgery.³ In this study, perioperative outcomes between robotics and VATS were not significantly different from each other.³ Another recent meta-analysis of 12 studies and 1173 patients with MG reported that VATS was associated with a reduction in post-operative adverse events (AEs), LOS, blood loss and myasthenic crises compared with open surgery; however, both groups had similar remission rates.⁵ Two other recent meta-analyses, which included a total of 20 studies and 1669 patients, found that subxiphoid thymectomies were associated with lower blood loss, pain scores and LOS than LVATS.^{4, 8} However, oncologic outcomes were either lacking or studies were limited to specific populations, such as patients with non-MG early-stage thymomas or MG only.^{4, 8}

There is limited information in the literature regarding the report of post-operative complications based on classification, type, severity and treatment efforts. Clinically, this presents challenges for deciding which surgical approach should be undertaken to optimise patient benefit. While existing studies are often retrospective and observational in nature, there remains a paucity of evidence and data to inform this controversial debate and consequently, further data are needed to inform best practices. To the best of our knowledge, no published study has simultaneously explored outcomes for robotics, subxiphoid, LVATS and open approaches for thymectomy. Our study aims to compare the safety and effectiveness of different surgical approaches in thymectomy—robotic, subxiphoid, LVATS and open—from a single Canadian centre.

2 METHODS

We performed a retrospective cohort study of patients who underwent thymectomy with a robotics, subxiphoid, LVATS, open sternotomy or thoracotomy approach for thymic lesions or MG at The Ottawa Hospital between July 1, 2017, and May 31, 2023. Exclusion criteria were age less than 18 years old and patients undergoing re-operation for recurrence. We identified 68 patients for inclusion in this study. The Modified Masaoka staging system and the World Health Organisation Classification of Tumours of the Thymus were used to determine the staging and histological subtype of thymomas, respectively.¹ The primary outcomes were operating time, estimated blood loss (EBL), conversion to open, intra-operative AEs, LOS, re-admission rates, 30- and 90-day mortality and AEs. Post-operative AEs were reported according to the Thoracic Morbidity and Mortality classification schema (https://ottawatmm.org/). Complications are divided into minor and major categories and grades: minor (I–II) and major (III–V).⁹ Grade I complications do not require any intervention; grade II complications require pharmacological treatment or intervention; grade III complications require surgical, radiological or endoscopic intervention, or multiple therapies; grade IV complications require intensive care unit care and life support and grade V complications lead to death.⁹ The operating time was defined as the time from skin incision to closure. Secondary outcomes were post-operative chest tube duration, resection margins, length of follow-up and oncologic recurrence.

3 RESULTS

A total of 145 potentially eligible cases were identified and examined for eligibility based on the study criteria and 68 cases were confirmed to be eligible for inclusion and analysis; following chart review, 77 cases were excluded because thymectomies were not performed, pathologies did not involve the thymus and there was missing pathology information or recurrence of disease.

No significant differences in age, sex, body mass index, Charlson comorbidity index, diffusing lung capacity for carbon monoxide and tumour size on CT at baseline between the robotics, subxiphoid, LVATS and open groups were found (Table 1). A greater proportion of patients with Eastern Cooperative Oncology Group (ECOG) status 1 was present in the subxiphoid group compared with the LVATS group (P < .05). Higher forced expiratory volume in 1 s (FEV₁) values were reported in the subxiphoid group (P = .018). Across all groups, thymectomies were mainly indicated for thymomas, thymic cysts and MG (Table 2).

Robotics, subxiphoid, LVATS and open approaches were used in 10%, 19%, 44% and 26%, respectively, of the total 68 thymectomy cases performed. Only 13% (4/30) of cases in the LVATS group were performed bilaterally. Left- and right-sided VATS were used in 12 and 14 cases, respectively. Of the seven robotic cases, four were completed using a right-sided approach, two cases were left sided and one case was bilateral.

Robotics, subxiphoid VATS and LVATS thymectomies were performed for modified Masaoka stages I or II thymomas (Table 3). Stage IV thymomas were only treated with open thymectomies. Patients with MG had ocular, bulbar symptoms, generalised MG or no symptoms. Patients with MG were receiving either pyridostigmine, prednisone or a combination. Plasmapheresis was given preoperatively to 65% (11/17) of patients with MG. No significant differences in operating time were found between the four groups (P = .166; Table 4). EBL was significantly lower in the LVATS groups compared with the open group (P < .001). Five planned LVATS cases and one planned robotics approach underwent conversion (two thoracotomies and four sternotomies, recorded as open thymectomies). Reasons for conversion were tumour adhesions, poor mobilisation and exposure. No conversions were observed in the subxiphoid group. No intra-operative AEs were seen across the robotics, subxiphoid and LVATS groups.

Chest tube duration was comparable across the four groups (P = .568). In the 34 patients with resected thymic tumours, the R0 resection rates were 100% in robotics, 83% in subxiphoid, 93% in LVATS and 75% in open cases. We had one case of thymic neuroendocrine tumour in the subxiphoid group which achieved R0 resection. One mucoepidermoid carcinoma (T1aNX in the LVATS group) and one thymic carcinoma (T1aN0 in the open group) also had R0 resection as well. Recurrence of thymoma at 15 months post-operatively was reported in one patient in the open group who had a Masaoka stage III thymoma at the first resection. The median hospital LOS was significantly shorter for robotics, subxiphoid and LVATS compared with the open group (P < .001). No deaths were reported in the robotics, subxiphoid or LVATS group. In the open group, we recorded one mortality each at 30 and 90 days post-operatively and one death 336 days post-operation. The last patient underwent a re-operation following a post-operative wound infection from the initial open thymectomy at almost 1 year. The remaining two deaths were caused by septic shock and heart failure. The highest count of post-operative AEs (n = 8) was reported in the open group. In both subxiphoid and LVATS groups, only minor complications (grades I and II) were observed. One grade IIIa complication was observed in one robotics case with increased drainage from a wound requiring sutures. One grade IIIb complication was seen in the open group (vocal cord injury requiring surgical intervention with general anaesthesia). Post-operative myasthenic symptoms (ocular only) were only reported in one case in the LVATS group.

4 DISCUSSION

Thymectomy is the treatment of choice for patients with thymic lesions and in certain cases of MG.¹ Different surgical techniques have evolved, including open median sternotomy or thoracotomy, LVATS, subxiphoid VATS and robotic-assisted thymectomy.^2-4 Conventionally, the open approach has been regarded as the gold standard and allows for direct visualisation of the thymus and adjacent structures. Minimally invasive approaches including LVATS have gained popularity in recent years due to growing evidence of lower post-operative pain, reduced LOS and comparable oncologic outcomes for thymic epithelial tumours.^{5-7, 10} While LVATS offers increased illumination and magnification of the operative field, it can be challenging to visualise the contralateral phrenic nerves.¹¹ In addition, intercostal nerve damage may occur as a result of intercostal incisions required for port placement.¹¹ By contrast, the subxiphoid approach allows for clear visualisation of the superior thymic poles and phrenic nerves bilaterally and avoids intercostal spaces altogether.¹¹ Advantages in robot-assisted thymectomies have also been described including greater stability and manoeuvrability of instruments within a small retrosternal area.¹² Currently, the literature surrounding the subxiphoid and robot-assisted approaches is limited.^{3, 8} A recent meta-analysis has evaluated subxiphoid to LVATS⁸ but comparisons between subxiphoid and the traditional open thymectomies have not been reported. The outcomes between subxiphoid and robotic thymectomies are also limited.

In our study, the operating time, defined as skin–skin, between robotics, subxiphoid, LVATS and open was similar, consistent with studies comparing LVATS with open^5-7 and robotics with open³ and robotics with LVATS.³ One meta-analysis revealed a shorter operating time for subxiphoid than LVATS.⁸ It is worthwhile to note that there is variability in the definition of operation time reported.³ Moreover, we found lower EBL in the LVATS group which may be attributed to the use of small incisions compared with an open sternotomy or thoracotomy. LVATS has been associated with reduced blood loss compared with open^5-7 and studies have also shown lower blood loss for subxiphoid versus lateral.⁸ While we expected shorter operating time and blood loss for subxiphoid VATS, differences may be attributed to variations in the learning curves between surgeons.

Here, we report a conversion rate of 7.4% (5/68) for LVATS, 1.5% (1/68) for robotics and 0% for subxiphoid. Meta-analyses have reported (only 11/13 studies had available data) an average conversion rate of 3.1% for LVATS⁷ and conversion rate ranging from 0% to 6.8%^{13, 14} for subxiphoid, citing bleeding, innominate vein injury and large vessel adhesion as major reasons.^14-16 Conversion rates for robotics range from 0% to 7.1%^{17, 18} and were mostly due to adhesions, tumour invasion into the innominate vein and poor tumour margins.⁴ This is consistent with the one robotics case in our study which was converted to open to optimise tumour margins. Despite having tumour adhesions (2/13), no subxiphoid cases underwent conversion in our study which is beneficial for patients as it avoids problems associated with open thymectomies including longer recovery, greater post-operative pain and potential for complications.

We observed a trend toward shorter mean chest tube duration for robotics (2 days), subxiphoid (2 days) and LVATS (2 days) compared with the open group (3 days). These lengths are generally shorter than what has been reported in the literature for the LVATS (average of 3.6 days), subxiphoid (median of 3.0 days) and open (average of 4.8 days) approaches.^{7, 19} One robotics study reported a shorter median chest tube length of 0 days compared with 3 days for open; in this study, it was routine to remove chest tubes before extubation to facilitate shorter LOS for robotics cases.²⁰ In our study, the robotics group was found to have the shortest LOS (1 day) followed by both subxiphoid VATS (2 days) and LVATS (2 days) than open (5 days). From the literature, LOS for robotics ranged from 1 to 9.6 days.^{18, 21} Our LOS for subxiphoid and LVATS is less than that reported in the current literature, averaging 3.7 days for subxiphoid⁸ and 6.2 days for LVATS.⁸ Improvement in LOS here may be explained by the implementation of an enhanced recovery after surgery program at our institution beginning in 2017.²² This program incorporates clinical pathways from the pre-operative to discharge period and provides designated patient education materials regarding perioperative expectations.²² The robotics group had the lowest incidence of post-operative complications followed by the subxiphoid VATS group in which all their complications were minor grade events (Grades I and II). Collectively, the shorter chest tube duration and LOS as well as decreased post-operative complications adds strength to the safety profile of robotics and subxiphoid VATS and highlights its potential to accelerate patient recovery and contribute to cost-savings for the hospital institutions.

Our median follow-up time was short for each cohort and thus, recurrence rates at 5 and 10 years were not evaluated in thymic epithelial tumours where complete resection is the gold standard. Tumour staging and completeness of resection are major factors influencing prognosis and risk of recurrence.¹ Higher rates of R0 resection were achieved in the robotics (100%), subxiphoid (83%) and LVATS (93%) groups than in the open group (75%), which may be explained by the fact that all thymomas in the minimally invasive cohorts were early stage (Masaoka stages I–II) and tumour size was smaller than that observed in the open group which had higher stage disease. R0 resection rates for a robotics approach ranged from 85% to 100%,^{3, 23-26} including one study that included cases of large thymomas (≥6 cm) and had an R0 rate of 95%.²⁵ To date, only a few studies have reported R0 resection rates for subxiphoid versus VATS; in a study comparing LVATS and uniportal subxiphoid, all cases had negative margins¹³ and in another study, R0 resection was 100% for both subxiphoid and LVATS.²⁷ These two studies only analysed thymomas of Masaoka stages I or II.^{13, 27} When comparing LVATS with open, R0 resection rates for thymomas and thymic carcinomas range from 96% to 100% for LVATS, and 70% to 96% for open.^{26, 28, 29} While we report high R0 resection rates for small and early-stage thymomas following subxiphoid, robotics and LVATS thymectomy, more studies are required to explore whether the same oncologic efficacy can be achieved for larger, more advanced thymomas.

Our present study has limitations. First, the interpretation of results between subxiphoid, LVATS and open thymectomy groups is limited by the small sample size and heterogeneity in clinical indications. This prevented us from conducting propensity score matching and multiple logistic regression to identify confounding variables. However, we found no differences in baseline characteristics between groups. Second, this study was non-randomised, retrospective and performed in a single institution, making it susceptible to selection bias. Third, we were unable to assess post-operative pain owing to reporting inconsistencies. Because of our short follow-up period, we could not assess long-term outcomes including overall survival, disease-free survival and remission rates for patients with MG. Larger prospective studies are needed to validate the findings observed in this retrospective cohort analysis. Lastly, we were unable to compare EBL in the robotics cohort with other cohorts as data were only reported qualitatively.

With regard to future directions, a larger sample size and multi-centred studies will be needed to validate R0 resection rates across the different thymectomy approaches discussed here. Further studies should explore long-term outcomes including overall survival, disease-free survival, recurrence for thymic epithelial tumours and remission rates for patients with MG following thymectomy.

To our knowledge, this is the first study to compare clinical outcomes between the newer robotics, subxiphoid, LVATS and open approaches and adds more data about thymectomy outcomes from a Canadian institution. Overall, our results show support for the less invasive robotics, subxiphoid VATS and LVATS as safe surgical approaches in thymectomy leading to shorter LOS compared with open median sternotomy or thoracotomy for thymic lesions and MG. The advantages of robotics, subxiphoid VATS and LVATS over open were shorter LOS, reduced post-operative complication rate and high rates of R0 resection.

CONFLICT OF INTEREST STATEMENT

None of the authors have conflicts of interest to declare.