Model predicts response to preoperative breast cancer chemotherapy
MedWire News: Researchers have devised a model to predict how breast cancer patients will respond to preoperative chemotherapy, based on the expression of biological markers including Ki-67, estrogen receptor (ER), and progesterone receptor (PgR) expression, as well as duration of preoperative therapy.
As pathologic response is considered a surrogate for long-term survival, the model may "provide early information on responsiveness of the disease," say Marco Colleoni (European Institute of Oncology, Milan, Italy) and colleagues in the European Journal of Cancer.
Recent studies have suggested that certain biological features might be able to predict a pathologic complete response (pCR) to preoperative chemotherapy.
However, Colleoni et al point out that these analyses are based on biological variables dichotomised as positive or negative, and no data are currently available on the continuous expression of biological features in the preoperative setting.
The researchers therefore set out to devise a sliding scale nomogram to assess the overall likelihood of a response for individual patients.
They recruited 783 patients with invasive ductal carcinoma who were treated with preoperative chemotherapy and operated on at the European Institute of Oncology.
All pathologic and clinical information was obtained at biopsy before initiation of therapy.
The researchers evaluated pCR according to Kuerer et al's criteria, namely the absence of invasive cancer on both the primary breast tumor and axillary lymph nodes.
On multivariable analysis, the probability of pCR was significantly associated with: Ki-67 expression, with an odds ratio (OR) of 1.15 for every 10% increase in the percentage of positive cells; the number of chemotherapy courses, with an OR of 1.31 per additional cycle; and inversely associated with ER and PgR expression, with respective ORs of 0.86 and 0.82 for every 10% increase in the percentage of positive cells.
The researchers constructed their nomogram for pCR based on these variables, and found it demonstrated good discrimination in the training cohort, as well as in a validation set of 38 patients - with area under the receiver operating characteristic curve values of 0.78 and 0.77, respectively.
"The quantitative assessment of biological features (eg, ER and PgR expression) may be important to better define treatment response, rather than simply relying on the positive and negative designations that encompass a mixture of patients with variable degree of responsiveness such as of endocrine responsiveness," Colleoni et al comment.
MedWire (www.medwire-news.md) is an independent clinical news service provided by Current Medicine Group, a trading division of Springer Healthcare Limited. © Springer Healthcare Ltd; 2010
By Andrew Czyzewski