Background: Nab-paclitaxel (albumin-bound paclitaxel) has shown preclinical activity against pediatric solid tumors. Preclinical data in EwS PDX models suggested high activity of nab-paclitaxel in tumors expressing high-levels of SPARC. Tumoral SPARC facilitates the accumulation of albumin in the tumor and increases the effectiveness of albumin-bound paclitaxel. Nab-paclitaxel utilizes albumin to deliver paclitaxel via caveolin-mediated endocytosis which is expressed in the EwS cells surface. We hypothesized that SPARC can be a predictive biomarker for nab-paclitaxel in EwS and DSRCT that could potentially be relevant for a better design of clinical trials and personalized treatments using nab-paclitaxel. Methods: Main endpoint of GEIS-39 was the overall response rate (ORR) assessed by RECIST 1.1 criteria with centralized pathology and imaging review. Secondary objectives included safety according to the CTCAE 4.0 criteria. Patients aged ≥ 6 months and ≤ 40 years, with relapsed/refractory DSRCT were eligible after having received at least one previous poly-chemotherapy line; EwS must have received at least two standard chemotherapy lines. Prior taxane therapy was accepted. Central pathology review selected for tumors with > Grade 3 (intense and diffuse) expression of SPARC by immunohistochemistry to be eligible. Nab-paclitaxel was administered as follows: age ≥ 21 and ≤ 40 years: 125 mg/m2 days 1, 8 and 15 in cycles of 28 days; age ≥ 6 months and ≤ 20 years: 240 mg/m2 days 1, 8 and 15 in cycles of 28 days. A 30% ORR was anticipated with a sample size of 25 patients needed to test the hypothesis. Stopping rule was set at 1 response within the first 16 treated pts. If 5 or more successes were observed in 25 pts, the results of the trial will warrant further investigation. Results: Twenty-nine patients were enrolled from June 2017 until October 2019, 11 DSRCT and 18 EwS. Median age was 32 years (range 14-69), and 5 females and 24 males were included, having received a median of 3 previous systemic treatment lines. Patients received a median of 3 cycles of nab-paclitaxel (range 1-17). In the EwS cohort an ORR of 33.3% (all partial responses, median duration 2 months) and 16.7% of stabilizations were achieved. No objective responses were observed among DSRCT pts, but 27.3% of pts achieved a stabilization. Overall, median progression free survival was 2.8 months and median overall survival 12.1 months, with no significant differences between DSRCT and EwS cohorts. Most common grade 3 toxicities were neutropenia (20.7%) and diarrhea (10.3%). Conclusions: Single agent nab-paclitaxel in biomarker selected EwS patients, but not in DSRCT, provided clinically meaningful activity that deserves further development. Nab-paclitaxel had a manageable adverse event profile. Clinical trial information: 2016-002464-14.