Textural characteristics of quantitative ultrasound spectral parametric maps have been proposed for the first time to predict cancer therapy response, early following treatment initiation. Such an early prediction can facilitate personalized medicine in cancer treatment procedures. Patients (n=10) with locally advanced breast cancer received neo-adjuvant chemotherapy, as "up-front" treatment, followed by mastectomy with axillary nodal clearance. Data collection consisted of acquiring tumor ultrasound radio-frequency data prior to neo-adjuvant treatment onset and at 4 times during treatment, in addition to pathological examinations of resected specimens after mastectomy. Several textural features were extracted from parametric maps of mid-band fit and 0-MHz intercept. The relative changes of these features were calculated one week after the treatment commenced, compared to the pre-treatment scan. Statistical analysis performed suggested that five of the applied textural features exhibit statistically significant differences between clinically/pathologically responding and non-responding patients. The promising results obtained represent a substantial step forward towards customizing cancer therapies by using this quantitative imaging modality. This can facilitate the switch of an ineffective treatment for a specific patient to a salvage therapy within weeks, instead of having patient endures months of the ineffective treatment.