Lymph node involvement may help predict metastases in breast cancer
Future development of distant metastases can be accurately predicted by combining primary tumour and stromal features with the histopathological patterns of involved and uninvolved lymph nodes (LN), a recent study has found.
Using the haematoxylin- and eosin-stained sections of 309 invasive breast carcinoma patients (median age 55 years), researchers characterized both the primary tumours and the axillary LNs. Multivariate proportional hazard regression was used to determine the sets of covariates that had the most accurate prediction of distant metastasis-free survival (DMFS).
Combining clinicopathological with histological features increases the accuracy in predicting 5-year DMFS. In all patients included, the combined covariates correctly predicted 5-year DMFS in 71 percent of the patients, as opposed to 68 and 62 percent of the patients when using only either set of features, respectively. In both sets of features, LN status and features were included.
In comparison, patients with triple-negative breast cancer (TNBC), histological features had higher accuracy than clinicopathological features (71 vs 67 percent) in predicting 5-year DMFS. The combined set of features still returned the highest accuracy.
In patients with LN involvement, incorporating five histological features (Salgado’s classification, size and number of germinal centres [GC] in uninvolved LN, location of GCs in involved LNs, and the presence of lymphocytic lobulitis) improved the predictive accuracy in all cancer patients and in TNBC patients.
Specifically, accuracy improved from 50 percent baseline performance to 64 percent in all breast cancer patients, and from 58 percent at baseline to 74 percent in TNBC patients.