MATERIALS AND METHODS
On behalf of the ESTS Thymic Working Group, emails were sent to ESTS members to ask for participation on this retrospective data collection in the treatment of patients with TETs with pleural involvement. Respondents received a detailed standardized questionnaire (Supplementary Material, Table S1). In this collaborative effort, a total of 152 patients with pleural disease of TETs were reported by 12 institutions from 8 countries. All institutions were specialized thoracic surgery centres routinely performing surgery on patients with TETs. The reported patients were diagnosed and operated between February 1977 and November 2014 (90% of patients between 1 January 2001 and 30 November 2014, Supplementary Material, Table S2).
Each institution obtained ethical approval from its ethics committee.

Study flow
There were coherent data for survival analysis on all 152 patients and on 115 patients with complete resection for analysis of freedom from recurrence (FFR; Fig. 1).
Figure 1  Flow chart illustrating the assignment of patients to different end-points. TETs: thymic epithelial tumours.

Statistics/outcome analysis
Definitions of recurrence and outcome measures were reported as recommended by the International Thymic Malignancies Interest Group [7]. Overall survival (OS) was calculated as the primary outcome from the date of surgery (first pleural surgery, Fig. 1) to the date of death of any cause. Patients without an event were censored at the last time point known to be alive. Disease-free survival was analysed from the date of surgery (first pleural surgery) to the date of recurrence or death from any cause (DFS) [8]. The end-point of interest for cause-specific survival (CSS) was defined as death from TET (censored observations: unrelated deaths and unknown cause of death) [7]. FFR was calculated only in patients after complete surgical resection (R0) from the date of surgery (first pleural surgery) to the date of recurrence and full information on recurrence status [9]. Log-rank tests were employed to compare whole Kaplan–Meier curves. Differences at 1, 3, 5 and 10 years were reported descriptively. Statistical methods are detailed further in the Supplementary Material.