2 Background: The OCM is a Centers for Medicare and Medicaid Services (CMS) alternative payment model, which seeks to curb costs while improving care for patients receiving systemic cancer therapy. CMS models the expected spending target for each 6-month episode using historical, geographic and clinical factors, including CTr participation. We evaluated the relationship between CTr participation, drug costs and performance in the OCM. Methods: We used claims for OCM episodes attributed to the Yale Cancer Center between July 2016 and July 2018. We stratified episodes by CTr participation and used t-tests and chi-square tests to compare total cost, drug costs (Part B, Part D and novel cancer therapies) and whether observed episode costs were above or below CMS targets. Analyses were conducted for the total sample and among the most common cancer types. Results: Among 9,387 OCM episodes (5,270 unique patients), 815 (8.7%) episodes involved a CTr. Among non-CTr patients, the mean Medicare cost per episode ($32,909) was modestly higher than the mean episode spending target ($31,746; p < 0.001), while in the CTr group, the mean Medicare cost per episode ($36,590) was substantially lower than the mean episode spending target ($48,124 p < 0.001). CTr episode costs were more likely to be under spending targets than non-CTr episodes (66% vs 56%, p < 0.001) overall and in breast, lung, and myeloma cancers, although only statistically significant for lung cancer (76% CTr vs 48% non-CTr, p < .001). Overall, non-CTr had significantly higher mean Part D drug costs per episode ($8,441 vs $3,893, p < 0.001), which was also noted among patients with lung cancer, ovarian cancer and lymphoma. Non-CTr episodes were also associated with higher mean novel therapies cost ($5,736) compared with CTr patient episodes ($4,346, p = 0.013). When comparing the sum of all other expenditures, CTr episodes were significantly associated with higher non-pharmaceutical expenditures than non-CTr episodes ($20,940 vs. $13,323, p < 0.001) overall. Conclusions: Episodes with CTr participation out-performed non-CTr episodes in achieving savings relative to CMS spending targets. Savings were driven by lower drug costs for the CTr episodes, particularly in the categories of Part D and novel cancer therapies. This suggests that CTr enrollment shifts costs for expensive pharmaceuticals away from CMS and toward the CTr study sponsor. Further research should explore whether this finding is generalizable to other cancer centers and payment models.