Systematic analysis of parameters predicting pathological axillary status (ypN0 vs. ypN+) in patients with breast cancer converting from cN+ to ycN0 through primary systemic therapy (PST)

• Published in: Clinical & experimental metastasis Vol. 35; no. 8; pp. 777 - 783
• Main Authors: Liedtke, C., Kolberg, Hans-Christian, Kerschke, L., Görlich, D., Bauerfeind, I., Fehm, T., Fleige, B., Helms, G., Lebeau, A., Stäbler, A., Schmatloch, S., Hausschild, M., Schwentner, L., von Minckwitz, Gunter, Loibl, S., Untch, M., Kühn, T.
• Format: Journal Article
• Language: English
• Published: Dordrecht Springer Netherlands 01.12.2018; Springer Nature B.V
• Subjects: Accuracy; Biomedical and Life Sciences; Biomedicine; Breast cancer; Cancer Research; Chemotherapy; Confidence intervals; Conversion; Estrogen receptors; Hematology; Lymph nodes; Model accuracy; Nodes; Nomograms; Oncology; Optimization; Parameters; Patients; Performance prediction; Regression analysis; Regression models; Research Paper; Statistical analysis; Surgical Oncology; Tumors; SENTINA trial; Primary systemic therapy; Prediction of lymph node status; Lymph node conversion
• ISSN: 0262-0898; 1573-7276
• DOI: 10.1007/s10585-018-9938-2

Optimization of axillary staging among patients converting from clinically node-positive disease to clinically node-negative disease through primary systemic therapy is needed. We aimed at developing a nomogram predicting the probability of positive axillary status after chemotherapy based on clinical/pathological parameters. Patients from study arm C of the SENTINA trial were included. Univariable/multivariable analyses were performed for 13 clinical/pathological parameters to predict a positive pathological axillary status after chemotherapy using logistic regression models. Odds ratios and 95%-confidence-intervals were reported. Model performance was assessed by leave-one-out cross-validation. Calculations were performed using the SAS Software (Version 9.4, SAS Institute Inc., Cary, NC, USA). 369 of 553 patients in Arm C were included in multivariable analysis. Stepwise backward variable selection based on a multivariable analysis resulted in a model including estrogen receptor (ER) status (odds ratio (OR) 3.916, 95% confidence interval (CI) 2.318–6.615, p < 0.001), multifocality (OR 2.106, 95% CI 1.203–3.689, p = 0.0092), lymphovascular invasion (OR 9.196, 95% CI 4.734–17.864, p < 0.001), and sonographic tumor diameter after PST (OR 1.034, 95% CI 1.010–1.059, p = 0.0051). When validated, our model demonstrated an accuracy of 70.2% using 0.5 as cut-point. An area under the curve of 0.81 was calculated. The use of individual parameters as predictors of lymph node status after chemotherapy resulted in an inferior accuracy. Our model was able to predict the probability of a positive axillary nodal status with a high accuracy. The use of individual parameters showed reduced predictive performance. Overall, tumor biology was the strongest parameter in our models.