A logistic regression model predicting high axillary tumour burden in early breast cancer patients

Abstract

Purpose

As elective axillary dissection is loosing ground for early breast cancer (BC) patients both in terms of prognostic and therapeutic power, there is a growing interest in predicting patients with (nodal) high tumour burden (HTB), especially after a positive sentinel node biopsy (SNB) because they would really benefit from further axillary intervention either by complete lymph-node dissection or axillary radiation therapy.

Methods/patients

Based on an analysis of 1254 BC patients in whom complete axillary clearance was performed, we devised a logistic regression (LR) model to predict those with HTB, as defined by the presence of three or more involved nodes with macrometastasis. This was accomplished through prior selection of every variable associated with HTB at univariate analysis.

Results

Only those variables shown as significant at the multivariate analysis were finally considered, namely tumour size, lymphovascular invasion and histological grade. A probability table was then built to calculate the chances of HTB from a cross-correlation of those three variables. As a suggestion, if we were to follow the rationale previously used in the micrometastasis trials, a threshold of about 10% risk of HTB could be considered under which no further axillary treatment is warranted.

Conclusions

Our LR model with its probability table can be used to define a subgroup of early BC patients suitable for axillary conservative procedures, either sparing completion lymph-node dissection or even SNB altogether.

http://ift.tt/2vTyFrk