Availability of patient-derived sarcoma models that closely mimic human tumors remains a significant gap in cancer research as these models may not recapitulate the spectrum of sarcoma heterogeneity seen in patients. To characterize patient-derived models for functional studies, we made proteomic comparisons with originating sarcomas representative of the three intrinsic subtypes by mass spectrometry. Human protein profiling was found to be retained with high fidelity in patient-derived models. Patient derived xenografts locally invade and colonize stroma in mice which enables unambiguous molecular discrimination of human proteins in the tumor from mouse proteins in the microenvironment. We characterized protein profiling of patient sarcoma tumors and mouse stroma by species-specific quantitative proteomics. We found that protein expression in mouse stroma was affected by the primary human tumor. Our results showed that levels of stromal proteins derived from the tumor were lowered in PDXs and cell lines and part of human stromal proteins were replaced by corresponding mouse proteins in PDXs. This suggests that the effects of the microenvironment on drug response may not reflect those in the primary tumor. This cross-species proteomic analysis in PDXs could potentially improve preclinical evaluation of treatment modalities and enhance the ability to predict clinical trial responses.