Radiologic Response to Neoadjuvant Treatment Predicts Histologic Response in Thymic Epithelial Tumors

Abstract

IntroductionNeoadjuvant treatment might increase resectability of thymic epithelial tumors (TETs). No standardized pathologic grading scheme for tumor response is available. Also, it is unclear whether radiologic treatment response can predict pathologic response. MethodsPatients with unresectable TETs who underwent neoadjuvant treatment before surgery at Mayo Clinic Rochester (1942–2014) were included. The pathologic tumor response grade (TRG) was based on Mandard grading (1994), ranging from TRG 1 (no viable tumor) to TRG 5 (no regression). TRG was compared with response by computed tomography, including with the Response Evaluation Criteria in Solid Tumors, version 1.1 (Byrne modification). ResultsA total of 49 patients, including 29 men, with a median age of 47.6 years and thymomas (n = 28) or thymic carcinomas (n = 21) were included. In five cases, pretreatment tumor type differed from posttreatment diagnosis. Thymic carcinomas had a greater morphologic response to neoadjuvant treatment than did thymomas with a lower percent viable tumor (p < 0.0001) and lower TRG (p<0.0001). Agreement for TRG by three reviewers was good (Krippendorff α = 0.838). By imaging (n = 24), partial response and larger reduction in tumor longest diameter and volume were associated with lower TRG (p = 0.0093, p = 0.0042, and p = 0.0021, respectively) and lower percent viable tumor (p = 0.0041, p = 0.0034, and p =0.0019). TRG correlated with radiologic change in tumor longest diameter and volume (Spearman correlation coefficient = 0.59 and 0.61, respectively). Radiologic change in tumor longest diameter and volume reasonably predicted pathologic TRG of 3 to 5 versus 1 or 2 (area under the curve 0.73 and 0.71, respectively). Sixty-seven percent of patients' tumors were completely resected. ConclusionsOur proposed histologic TRG for TETs appears easy and reproducible and correlates with radiologic response. Radiologic response is useful to predict pathologic response.