Several previous reports suggest that thrombospondin (TSP-1) may be a mediator of the antiangiogenic effects of low-dose metronomic cyclophosphamide-based chemotherapy (MC). We conducted a randomized phase II trial evaluating megestrol acetate (n = 44) versus MC (n = 44) in patients having exhausted all standard treatments. We measured the TSP-1 levels at baseline and D15. We did not observe significant differences in TSP-1 at baseline in the two arms (p = 0.07). TSP-1 levels decreased in patients receiving metronomic cyclophosphamide (from 16.6 ± 7.2 µg/ml to 12.8 ± 7.4 µg/ml; p = 0.057). The TSP-1 level was stable in patients receiving megestrol acetate. Nevertheless, the TSP-1 level driven by MC did not correlate to clinical benefit.