Abstract
BACKGROUND: Tumor-associated collagen signature (TACS) is an independent prognostic factor for breast cancer. However, it is unclear whether the complete collagen signature, including TACS, the TACS-based collagen microscopic features (TCMF1), and the TACS-based nuclear features (TCMF2), can provide additional prognostic information for the current tumor-node-metastasis (TNM) staging system. PATIENTS AND METHODS: We included 941 patients with breast cancer from three cohorts: the training (n = 355), internal (n = 334), and external validation cohorts (n = 252). TACS and TCMF1 were obtained by multiphoton microscopy (MPM). TCMF2 was extracted on the hematoxylin and eosin images colocated with MPM images. They were linearly combined to establish a complete collagen signature score for reclassifying current TNM staging into stage Ⅰ (II and Ⅲ)/low risk and stage Ⅰ (II and Ⅲ)/high risk. RESULTS: The low-risk collagen signatures 'downstaged' patients in stage II or Ⅲ, while the high-risk collagen signatures 'upstaged' patients with stage Ⅰ tumors. After incorporating the complete collagen signature into the current TNM staging system, the modified staging system had a higher ability to stratify patients [referent, Ⅰ-new; Ⅱ-new, hazard ratio (HR) 8.655, 6.136, and 4.699 in the training, internal validation, and external validation cohorts, respectively; Ⅲ-new, HR 14.855, 11.201, and 13.245 in the corresponding three cohorts, respectively] than the current TNM staging system (referent, Ⅰ; Ⅱ, HR 1.642, 1.853, and 1.371 in the corresponding three cohorts, respectively; Ⅲ, HR 4.131, 4.283, and 3.711 in the corresponding three cohorts, respectively). Furthermore, the modified staging system showed a higher area under the curve than the current TNM staging system (training cohort: 0.843 versus 0.683; internal validation cohort: 0.792 versus 0.661; and external validation cohort: 0.793 versus 0.646). CONCLUSIONS: The complete collagen signature is an independent predictor of survival outcomes in breast cancer. It adds significant information about the biological behavior of the disease to staging for breast cancer.