Abstract 148P
Background
Metabolomics studies metabolites in biological samples. Cancer can impair metabolism, so the pattern of altered metabolites could reproduce a “cancer signature”. This analysis aimed at identifying a “metabolic signature” that could differentiate elderly eBC patients (pts) from elderly advanced breast cancer (aBC) pts, as well as at investigating its prognostic role in terms of disease recurrence (DR).
Methods
Serum samples from elderly BC pts enrolled in 3 onco-geriatric trials coordinated by the Medical Oncology Division of Prato, were retrospectively analyzed via proton nuclear magnetic resonance (NMR) spectroscopy. Three NMR spectra were acquired for each serum sample (NOESY1D, CPMG and Diffusion-edited). Random Forest (RF) models were calculated. The ability of metabolomics to predict BC relapses was assessed using Kaplan–Meier curves with calculation of the hazard ratio (HR) and p-value by Log-Rank test.
Results
Serum samples from 140 pts with eBC and 27 with aBC were collected between 2008 and 2018. In the aBC cohort, median age was 79 years (95% CI, 70-88); 25.9% of pts were hormone receptor positive HER2 negative (HR+HER2-), 29.6% HER2 positive (HER2+), 11.1% triple negative (TN); data missing in 33.3% of cases. In the eBC cohort, median age was 76 years (95% CI, 70-91); 30.7% of pts were HR+HER2-, 48.6% HER2+, 11.4% TN; data missing in 3.6% of cases. In this cohort, 39.3% of pts had tumors larger than 2 cm, and 41.4% had positive axillary lymph nodes. Using NOESY1D spectra, the RF classifier discriminated free from recurrance eBC from aBC with a sensitivity, specificity and accuracy of 81.5%, 66.7% and 70% respectively, performing better than the other spectra. Therefore, we tested the NOESY1D spectra of each eBC pt on the RF models already calculated. If the RF model classified the sample as relapsed, it was considered at “high risk” of DR. Our analysis showed that pts classified as "high risk” had a higher risk of DR (HR 3.39, 95% CI 1.59-7.24, p=0.00084).
Conclusions
This analysis suggests that a “metabolic signature”, identified employing NMR spectral profiling, is able to predict the risk of DR in elderly pts with eBC. Further studies are needed to confirm these data.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
Azienda USL Toscana Centro.
Funding
\"Sandro Pitigliani” Foundation.
Disclosure
All authors have declared no conflicts of interest.