P2-17-07: Construction of a Predictive Model of Probability of Ovarian Function Recovery in a Series of Premenopausal Breast Cancer Patients with Chemotherapy-Induced Amenorrhea Switched to an Aromatase Inhibitor (AI) after Adjuvant Tamoxifen.

Abstract

Abstract Introduction: AI therapy is not recommended in breast cancer patients with conserved ovarian function as AI decreases estrogenic feedback leading to an increased of FSH and LH. In patients older than 40 and with prolonged amenorrhea, switching to an AI after tamoxifen therapy is a controversial approach but with an adequate follow-up for early detection of potential menses renewal it might be feasible. Several factors have been identified in the literature as clinical defining variables of CIA while other factors have been associated with high probability of permanent CIA. Among clinical defining variables are time from last menses date (LMD) and low estradiol (E2) levels. Risk variables for permanent CIA are chemotherapy schedule administered and advanced age. We aimed to construct a predictive model to identify high risk of renewal of menses after switching to AI in premenopausal patients with CIA lasting at least 1 year. Methods: Based on defining and prognostic variables of CIA, a predictive model of high probability of permanent amenorrhea was constructed by assigning a score as follows: 1) Time from LMD to switching date >3 years: 1 point, < 3 years: 0 points 2) E2 levels <20 ng/ml: 1 point, >20 ng/ml or unknown: 0 points 3) Age > 45 years: 1 point, < or = 45: 0 points 4) Chemotherapy regimen administered: dose dense or high doses: 2 points, conventional doses regimen with anthracycline and taxanes: 1 point, conventional doses only anthracycline-based or other: 0 points. Final score obtained from 0 to 5 points was classified in two groups: Low-probability of permanent amenorrhea: score 0–2, and high probability of permanent amenorrhea: score 3–5. To validate this probability model we retrospectively analysed data from a prospective maintained database of early breast cancer patients, clinically premenopausal at diagnosis that were treated in our institution from May 2004 to December 2009. All patients had histologically confirmation of hormone-sensitive breast carcinoma at stage I-III. Therapy included in all cases adjuvant or neoadjuvant chemotherapy and adjuvant tamoxifen and all patients referred a CIA lasting at least 1 year. Since 2006, in most patients E2 levels before switching were determined. Results: Validation of probability model was performed in our series of 102 premenopausal patients of whom 9 recovered ovarian function (prevalence ratio: 0.088). Test sensibility of the model was 100% and specificity 56.99%. Positive predictive value was 18.4% and negative predictive value 100%. Positive likelihood ratio was 2.32 and negative likelihood ratio was 0. Conclusions: Our model of probability of ovarian function recovery is a highly sensitive test that might become a useful tool to identify premenopausal patients with CIA that could be safely switched to AI after tamoxifen. Further validation in a prospective series including more accurate E2 monitoring and follow-up is warranted. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P2-17-07.