THE ITMIG CLASSIFICATION OF THYMOMA RECURRENCE: SURVIVAL ANALYSIS AND PERSPECTIVES.

Abstract

OBJECTIVE\nITMIG classifies thymoma recurrences based on the topographic location, but its effectiveness in prognosis prediction has not been well investigated yet. Aims of this study are to analyse survival outcome of patients surgically treated for thymoma recurrence according to the ITMIG recurrence classification and to investigate possible alternatives.\n\n\nMETHODS\nFrom 01/01/1990 to 1/07/2017 data on 135 surgically treated patients for thymoma recurrence from three high volume centres were collected and retrospectively analysed. Patients were classified according to the ITMIG classification as local, regional and distant.The ITMIG classification and alternative classifications were correlated to overall survival(OS).\n\n\nRESULTS\nAccording to the ITMIG classification, recurrence was local in 17(12.5%), regional in 97(71.8%) and distant in 21(15.7%)patients, with single localization in 38(28.2%) and multiple localizations in 97(71.8%). Five- and ten-year OS were 79.9%and 49.7%in local, 68.3%and 52.6%in regional, 66.3%and 35.4%in distant recurrences, but differences were not statistically significant(p=0.625). A significant difference in survival was present considering single vs multiple localizations: 5 and 10 yearsOS of 86.2%and 81.2% vs 61.3% and 31.5% (p=0.005,HR7.22,95%CI=0.147-0.740). Combining the localization number with the recurrence site, ITMIG loco-regional single recurrence had a statistically significant better survival compared to patients with ITMIG loco-regional multiple recurrence or ITMIG distant recurrence(p=0.028). Similarly, a significant difference was present considering intrathoracic single vs intrathoracic multiple vs distant recurrence(p=0.024).\n\n\nCONCLUSION\nThe ITMIG classification for thymoma recurrence did not show significant survival differences comparing local, regional and distant recurrences. Integrating this classification with the number of the localizations may improve its effectiveness in prognosis prediction.