Stromal Responses among Carcinomas—Letter

Chen and colleagues (1) described a fibroblastic signature desmoid-type fibromatosis (DTF) as a common feature in different types of carcinomas (prostate, ovarian, colon, and lung). These data are very interesting and relevant from a biologic point of view because they suggest that for a subgroup of patients, this feature represents a common stromal response to cancer in different organs, but with opposite prognostic power according to tumor type. Indeed, although DTF predicts worse outcome in ovarian and colon carcinomas, the same authors found association of this signature with good outcome in breast carcinomas (2). However, the authors' conclusion is not fully warranted on the basis of the findings presented.

Stromal signatures discovered in breast carcinomas, including stromal metagene (3), SDPP (4), and our own ECM3 (5), have higher or even opposite prognostic/predictive power when combined with existing outcome predictors [i.e, estrogen receptor (ER), HER2, and grade]. We found that ECM3, characterized by upmodulation of DTF-associated genes (6), has an independent negative prognostic role only in chemotherapy-untreated grade 3 breast carcinomas, whereas it is slightly associated with better outcome in differentiated tumors (5). Thus, even in absence of a significant association between DTF and specific tumor characteristics, this stromal signature may influence disease progression differently in the same tumor type according to characteristics of the neoplastic cells. Although the disparity in prognostic results according to tumor type is attributable to variations in carcinoma stage in the publicly available datasets, as postulated by the authors themselves, it also raises the possibility that DTF, conserved across different types of carcinomas, is associated with the same responses in tumors of different origin but with similar stage characteristics. In addition, because chemotherapy significantly affects the features of the tumor microenvironment (7), analysis of the clinical significance of DTF should be performed separately in tumors of treated and untreated patients. Accordingly, stromal metagene, which is strongly correlated with DTF (2), was found predictive in ER-negative chemotherapy-treated carcinomas but not prognostic in untreated ER-negative breast carcinomas (3). In summary, we suggest that conclusions about a clinical role for DTF in different tumor types await further analyses that consider the interaction of tumor stromal characteristics with neoplastic cell molecular traits and with the therapeutic treatment of the patient.

See the Response, p. 1397