|Home | About | Journals | Submit | Contact Us | Français|
Minimally invasive thymectomy (MIT) is a surgical approach to thymectomy that has more favorable short-term outcomes than open thymectomy (OT) for myasthenia gravis. When performed for thymoma, the oncologic outcomes of MIT have not been rigorously evaluated. We analyzed determinants of complete (R0) resection among patients undergoing MIT and open thymectomy in a large international database.
The retrospective database of the International Thymic Malignancy Interest Group (ITMIG) was queried. Chi-Square and Wilcoxon rank-sum tests, multivariate logistic regression models, and propensity matching were performed.
A total of 2514 patients underwent thymectomy for thymoma between 1997 and 2012. 2053 (82%) patients underwent OT, 461 (18%) patients underwent MIT, and the use of MIT increased significantly in recent years. The rate of R0 resection among patients undergoing OT was 86%, and among those undergoing MIT was 94%, respectively (p<0.0001). In propensity matched MIT and OT groups (n=266 each group), however, the rate of R0 resection did not differ significantly (MIT 96%, OT 96%, p=0.7). Multivariate analyses were performed to identify determinants of complete resection. Factors independently associated with R0 resection were geographical region, later time period, less advanced Masaoka stage, total thymectomy, and the absence of radiotherapy. Surgical approach, whether minimally invasive or open, was not associated with completeness of resection.
The use of MIT for resection of thymoma is increasing substantially over time, and MIT can achieve similar rates of R0 resection for thymoma as OT.
Minimally invasive surgical approaches for thymectomy have gained popularity for patients with non-thymomatous myasthenia gravis (MG) and a variety of single institution studies suggest that that minimally invasive thymectomy (MIT) has improved short-term outcomes when compared with open thymectomy (OT). For example, when compared with OT, MIT has been associated with shorter hospital stays 1–7, shorter length of stay in the intensive care unit 1, lower operative blood loss 1, 7, 8, improved postoperative pulmonary function 9, decreased postoperative pain 10, and with similar operating room times 1.
Currently, there are few published oncologic outcomes data that support the adoption of MIT for patients with thymic malignancies. For resection of thymoma, MIT and OT could vary with respect to rates of complete resection, extent of thymectomy, and risk of violation of the tumor capsule. It is well known that complete resection is the most important long-term prognostic indicator in thymoma, and resection status at the time of thymectomy closely correlates with recurrence outcomes 11–15. Using the large international database of the International Thymic Malignancies Interest Group (ITMIG), we therefore set out to: 1) determine whether MIT was associated with equivalent rates of incomplete resection compared with OT for thymoma, and 2) identify factors that were independently associated with completeness of resection.
The ITMIG database is a multi-institutional retrospective database with contributions from 43 institutions on 4 continents 16. All data were de-identified and considered exempt from IRB review. This database was queried to define a cohort of patients who underwent OT or MIT for thymoma, from the date of the first MIT entry in 01/1997 up until the date of last entry in 12/2012. This study was conducted on the data available in the ITMIG database at the time of data harvest. For this project, OT was defined as either sternotomy or thoracotomy, and MIT was defined as either a video-assisted thoracoscopic surgical (VATS) approach or a robotic-assisted thoracoscopic surgical approach (RATS). Thoracosternotomy and clamshell approaches, cervical thymectomy, and mediastinotomy approaches were intentionally excluded as they represented a minority of resection types, and to balance the OT and MIT comparison groups. Variables examined in our analyses are listed in Table 1 and each contained less than 5% missing data. The clinical stage variable was intentionally excluded because of 28% missing data. A total thymectomy was defined in the database as resection of the thymoma and the entire thymus gland whereas a partial thymectomy was defined as resection of the thymoma with a margin of normal thymic tissue. The R0, R1, and R2 variables were defined in the database as being identified from surgical pathology reports. Conversion from MIT to OT data was not available from the retrospective ITMIG database.
All statistical analyses were performed with SAS 9.3 (SAS Institute, Cary, NC). Patient characteristics were compared between MIT and OT groups using the chi-square test for categorical variables and the Wilcoxon rank sum test for continuous variables. Propensity matching was performed by nearest available pair matching method, utilizing all of the variables in this study with the exception of completeness of resection. The MIT patients were ordered and sequentially matched to the nearest unmatched OT patients. If multiple OT patients have propensity scores that are equally close to that of the MIT patient, one of these OT patients is selected at random. A p<0.05 was used to determine significance. Univariate logistic regression analyses were used to investigate the relationship between resection status and clinical variables and patient characteristics variables including age, gender, paraneoplastic syndrome, time period, WHO histology, pathologic Masaoka stage, tumor size, extent of thymectomy, surgical approach, use of chemotherapy and radiotherapy. A multivariate logistic regression model was built to test the association between resection status and surgical approach (MIT vs OT), including other clinical factors and patients characteristics variables as covariates.
A total of 2514 patients underwent thymectomy for thymoma between January 1997 to December 2012. 2053 (82%) patients underwent OT [1751 by sternotomy (85%) and 302 by thoracotomy (15%)], and 461 (18%) patients underwent MIT [315 by VATS (68%) and 146 by RATS (32%)]. The number of MIT and OT cases performed from 1997 to 2012 is shown graphically in Figure 1. The frequency of MIT cases was seen to increase from 2008 to 2012 where it approached the frequency of OT cases.
The demographic, clinical, and pathologic characteristics of these cases are shown in Table 1. The majority of MIT cases were contributed from Asian institutions. Patients undergoing MIT were 2.1 years younger than those undergoing an OT, and patients in the MIT group were more likely to be female (55%) than those in the OT group (45%). MIT patients had more favorable World Health Organization (WHO) histology and less advanced pathologic Masaoka stage than patients undergoing OT. In the MIT group, the mean tumor size was 4.0 cm (range 0.4 to 15.5 cm), compared with 6.0 cm (range 0.1 to 28.0 cm) in the OT group. The rate of partial thymectomy was found to be higher in the MIT group (27%) than in the OT group (9%). Patients undergoing MIT were less likely to have received chemotherapy (3%) or radiotherapy (31%) than those undergoing OT (16% and 37%, respectively), and the majority of chemotherapy and radiotherapy were administered in the adjuvant setting.
Eighty-eight percent of all thymectomies achieved complete (R0) resection, and R0 resection was achieved more frequently in patients undergoing MIT (94%) compared with OT (86%). There was a statistically significant relationship with R0 resection and more recent time period for all patients (p<0.0001), and Figure 2 shows these relationships in the MIT (p=0.22) and OT cohorts (p=0.02). Among MIT resections, the rate of R0 resection in VATS thymectomy was 95%, and in RATS thymectomy was 92% (p=0.2). To determine whether completeness of resection was independently associated with surgical approach, propensity matching was performed between MIT and OT groups. As shown in Table 2, MIT and OT groups each contained 266 patients and were balanced with respect to all variables. The rate of R0 resection did not differ between balanced MIT and OT groups, where it was 96% in each group (p=0.7). A sensitivity analysis of patients with stage I–II thymoma was performed in 385 patients undergoing MIT and 1313 patients undergoing OT. This analysis demonstrated no significant difference in R0 resection between patients in the MIT group (96.6%) and OT group (95.1%) (p=0.19).
To examine which factors were associated with R0 thymoma resection, univariate analyses were performed in all patients undergoing thymectomy (MIT and OT). Variables significantly associated with R0 resection included geographic region, later time period, female gender, less favorable WHO histology, less advanced Masaoka stage, total thymectomy, minimally invasive surgical approach, and not receiving chemotherapy or radiotherapy (Table 3). Multivariate analyses were next performed to determine which of these factors were independently associated with R0 resection status. In these analyses, surgical approach, minimally invasive or open, was not a predictor of R0 resection. Variables that were independently associated with R0 resection included region, more recent time period, less advanced Masaoka stage, total thymectomy, and not receiving radiotherapy (Table 4).
To advance the knowledge of thymic malignancies, the International Thymic Malignancy Interest Group (ITMIG) has developed a centralized, international, retrospective database, which currently represents the largest dataset assembled for these relatively infrequent tumors 16. Because the oncologic effectiveness of MIT is essentially unknown, we utilized this dataset to determine whether thymectomy performed by minimally invasive techniques has equivalent rates of complete thymoma resection as thymectomy performed by sternotomy or thoracotomy.
Completeness of resection is a critically important oncologic outcome that is highly correlated with disease recurrence in thymoma 11–15. The study of disease recurrence itself can be challenging to evaluate following thymoma resection primarily because of the long (approximately 10 year) disease-free intervals seen after thymectomy for thymoma. It is therefore not surprising that the several studies that evaluated survival following minimally invasive resection of thymoma have relatively short follow-up periods that range from 24 to 40 months 1, 2, 8, 17, which are too truncated to draw definitive conclusions about survival and relapse following surgical resection of thymoma. Similarly, at the time of data harvest for our study, the number of recurrence events were insufficient to effectively study recurrence as a primary outcome variable.
The performance of MIT and OT techniques with regard to rates of complete (R0) resection has not been rigorously evaluated. In one previous single institution study of 10 patients undergoing MIT for thymoma, MIT was associated with a 100% rate of R0 resection 1. In our study, R0 resection was achieved in 94% of all MIT (n=461) and 86% of all OT procedures (n=2053, p<0.0001). Whereas this difference was likely associated with allocation of more advanced thymoma tumors to the OT group, our propensity matched analyses of balanced MIT and OT groups demonstrated equal rates of R0 resection in MIT (96%) and OT groups (96%) (p=0.7). Further, in multivariate logistic regression analyses, surgical approach was not associated with completeness of resection. Taken together, this data supports equivalence in the ability to achieve R0 thymoma resection by minimally invasive and open surgical approaches.
Variables that were independently associated with R0 resection in our study included less advanced Masaoka stage, the absence of receiving any radiotherapy, more recent time period, the geographic region on which the thymectomy was performed, and total (versus partial) thymectomy. The relationship of Masaoka stage and of radiotherapy to completeness of resection can be reasonably ascribed to invasiveness of the tumor. Further, it is likely that the relationship of more recent time period to completeness of resection, especially for MIT, can be attributed to evolving surgical technique. The finding that the rate of R0 resection was significantly lower in the Americas (which was predominantly North America-North America 98% and South America 2%) compared with other continents was interesting. Of all cases of thymectomy in North America, R0 resection rates were 78%, which was lower when compared with thymectomies performed in Asia (92%) and Europe (91%). Whereas this could possibly be due to a higher frequency of invasive thymoma resections being performed in North America compared with other continents, thymectomy performed in North America was independently associated with R1/R2 resection in multivariate analyses, and the reasons for this are not known.
Among all thymecotomies this study, which included early and advanced stage tumors, total thymectomy (total resection of the thymus gland) was associated with complete (R0) resection of the thymoma. Whereas some authors recommend total thymectomy for all cases of thymoma resection including those cases involving only partial involvement of the thymus 18, there are few objective data to substantiate the assumption that this practice will result in more favorable cancer-specific outcomes, and it is certainly technically feasible to achieve R0 tumor resections of smaller thymomas by partial thymectomy. For example, one report of 61 patients undergoing thymectomy for thymoma demonstrated better overall survival after complete thymectomy compared resection of the tumor alone 19, although another study on 126 thymoma resections suggested that there was no difference between total and partial resection of the gland 14.
In this study, 27% of patients in the MIT group underwent partial thymectomy compared with 9% of patients in the OT group. This could possibly be related to the unilateral surgical approaches that are most often performed in MIT, compared with the “bilateral” approach afforded by sternotomy. Until further data become available, we support total thymectomy for patients undergoing resection for thymoma. In patients with a thymoma who also have MG, an “extended” thymectomy can be considered where the entire gland is removed along with the contiguous right and left mediastinal pleura, mediastinal and pericardiophrenic adipose tissues, and dissection of the aorto-pulmonary window 20. Unencapsulated lobules of thymus and microscopic foci of thymus can be widely and invisibly distributed in the pretracheal and anterior mediastinal fat, and a number of studies have suggested that removal of more thymic tissues is associated with improved remission rates for MG 21.
Proponents of open approaches to thymectomy discuss excellent visualization of the thymus and its surrounding structures as well as the ease of resection and reconstruction of involved structures in cases of advanced disease. A similar perspective would view the open approach as providing the most oncologically sound operation, decreasing the risk of capsule violation, and in cases of MG, facilitating removal of the entire thymus gland and associated mediastinal fat (25). Proponents of MIT endorse improved visualization of the thymus gland and surrounding structures over OT, in particular with regard to the magnification and advanced optics of current robotic technology, and will cite improved and shorter recovery following surgery.
A number of studies on thymectomy, predominantly for MG, have demonstrated favorable short-term advantages of MIT when compared with OT. These advantages include improved pain scores, shorter recovery, decreased blood loss, and shorter length of hospital and ICU stays 1–10. In our study, in patients undergoing MIT, 68% underwent a thoracoscopic approach and 32% underwent a robotic approach. Whereas many of the initial reports of MIT included thoracoscopic approaches, more contemporary series have evaluated both thoracoscopic and robotic techniques. The balance of these reports suggest that robotic assisted approaches, albeit performed mostly for MG, are safe and can result in good short-term outcomes 22. The largest of these studies is a multi-institutional retrospective study of 79 robotic assisted thymectomies from 4 European centers that reported a median hospital stay of 3 days and a 13% complication rate 17. Similarly, in a retrospective comparison of 79 thoracoscopic with 74 robotic thymectomies for MG, there were similarly low morbidity rates and rates of conversion between these two minimally invasive approaches 23.
Certainly not all patients with a thymoma are suitable candidates for MIT. Suggested exclusion criteria for MIT include involvement of a phrenic nerve, innominate vein, or other major vessels 20. Based upon a retrospective review of MIT performed with R0 resection of thymoma, one group has suggested that selection criteria for MIT include location of the tumor in the mediastinum, distinct fat planes between tumor and vital organs, tumor encapsulation, the existence of residual normal appearing thymic tissue, no mass compression effect, and unilateral tumor predominance, 24. Bilateral MIT approaches, however, are well described and tumors without unilateral predominance can be approached in such a way. In our dataset, the median tumor size resected by MIT was 5.5 cm (range 0.4 to 15.5 cm), and tumor size was not predictive of R0 resection, suggesting that size of thymoma is not as an important consideration for the decision to proceed with MIT as are other factors representing tumor invasion. Generally, patients with a high suspicion for invasion of mediastinal structures are better served by open thymectomy. Further, should minimally invasive resection at any time compromise basic oncologic principles, conversion to OT is necessary.
The strengths of our study include the large size of the patient cohort and the multi-institutional, multinational database from which it was derived. Limitations of this study include its retrospective nature and associated selection and information biases. This study was additionally limited by some certain missing data, for example the conversion of MIT to OT. These instances were likely coded as OT and could bias the R0 resection rate against OT. For reference, such published conversion rates are 0–7% (1,8,17). Additionally, certain approaches to thymectomy (thoracosternotomy clamshell, and cervical) were excluded from our analyses secondary to their relative rarity in our database. Certainly these approaches may be warranted and beneficial in certain cases. Indications for thymectomy are not available in the retrospective ITMIG database, and it is possible that smaller tumors were incidental findings of thymectomy for MG or other neoplastic syndrome. Lastly, although this is a large multi-institutional international database comprised of select participants, it is difficult to precisely estimate its representation of the general population in the U.S. and the world.
Our data demonstrates that worldwide adoption of MIT for thymoma has increased substantially in recent years. In our dataset comprised of data from 43 institutions in 15 countries, approximately half of all thymectomies performed in recent years for thymoma were performed by a minimally invasive technique. Our data demonstrate that R0 resection of thymoma can be performed at comparable rates by minimally invasive and open thymectomy operations. The long-term cancer-specific outcomes of MIT will be determined in future studies with maturation of the ITMIG database.
Sources of Funding: none
The authors would like to thank Frank Detterbeck for critical review of data, appraisal of the manuscript, expert insight, and support for this project. The authors would also like to thank Elham Biaghoshi for administrative support for this manuscript and project.
Disclosures: No authors have a conflict of interest
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.