Nodal stage classification for breast carcinoma: improving interobserver reproducibility through standardized histologic criteria and image-based training.

PURPOSE: Reliable pathologic stage classification of axillary lymph nodes is an important determinant of prognosis and therapeutic decision making for patients with invasive breast cancer. Pathologists' distinction between micrometastasis (pN1mi) and isolated tumor cells [ITC; pN0(i+)] is variable using the American Joint Committee on Cancer (AJCC) Staging Manual (Sixth Edition). We sought to determine whether a set of clearly defined histologic criteria could lead to reproducible nodal classification by pathologists. PATIENTS AND METHODS: Digital images of sentinel lymph node biopsies from 56 patients with small-volume nodal metastases were examined by six experienced breast pathologists (MDs), first as a pre-test, and again as a post-test after studying a training program that outlined and illustrated the classification criteria.
RESULTS: Post-test results, after study of the training program, were significantly improved. Compared with the reference MD, agreement improved from 76.2% (pre-test kappa = 0.575; standard deviation [SD], 0.25) to 97.3% (post-test kappa = 0.947; SD, 0.049). Multirater analysis of agreement among the six MDs improved from 71.5% (pre-test kappa = 0.487; ASE, 0.039) to 95.7% (post-test kappa = 0.915; ASE, 0.037). Agreement on lobular carcinoma metastasis classification improved from 55% (23 of 42; pre-test) to 100% (42 of 42; post-test) (P < .001), and agreement on ITC classification in nodal parenchyma improved from 67.6% (69 of 102; pre-test) to 98.0% (100 of 102; post-test; P < .001).
CONCLUSION: Application of current definitions for classification of small-volume nodal metastases are inconsistent, leading to variable classification of ITC and micrometastases. Reproducibility of pathologic nodal stage classification is achievable through study of a training set to clarify the AJCC criteria.