Retrospective Study of 2nd-line Therapies After CDK4/6i + Hormonal Therapy in HR+/HER2- Advanced Breast Cancer

Multicenter retrospective observational study, describing therapeutic choices as second-line treatment in patients with HR + / HER2- advanced breast cancer in a real world setting, in centers adhering to the Hermione Network.

The study involves the analysis of patients with HR + / HER2- advanced breast cancer treated in second line after initial treatment failure with aromatase inhibitor or Fulvestrant + CDK4 / 6i (Palbociclib, Ribociclib or Abemaciclib).

Data will be collected from 150 patients with at least one radiological re-evaluation of disease during 2nd-line treatment from 01 January 2016 until 31 December 2020.

List of collected information: Patients' characteristics (gender, age at diagnosis, menopausal state); Disease definition at diagnosis (stage, tumour histology, hormonal status); Surgery (date of surgery, type of surgical approach); Neo-adjuvant treatment; Adjuvant treatment; Date of first relapse (and time since the end of adjuvant therapy); Locations of metastases, Biopsy of metastases, Hormonal receptor status; First-line treatment, hormonal therapy, Best response, Cause of treatment end; Second-line treatment, Best response (radiological re-evaluation), Toxicity, Cause of treatment end.

Demographics, baseline characteristics (including tumor characteristics) and treatment information will be summarized descriptively. The categorical variables will be presented in the form of frequencies and percentages, while the continuous variables will be presented by mean, standard deviation and minimum and maximum values.

A logistic model will be used for the analysis of clinical benefit (categorical variable), while for the analysis of time-to-event indicators a proportional hazard model will be used. For both analyses, the optimal model will be chosen with the method of "backward" selection. A threshold value of 5% will be used to include predictive variables in the model. The estimates derived by the final models will be evaluated using a penalized model for the evaluation of maximal probability, according to the Firth approach.