Background: Polyclonal emergence of KIT secondary mutations (muts) is the main mechanism of imatinib (IM) progression in GIST. Although approved KIT inhibitors SU and RE each suppress only a subset of these muts, they have shown complementary activity in GIST models and clinical trial correlates. Preclinical evidence suggests that rapid alternation of SU and RE broadens the spectrum of IM-resistant subclones targeted, compared to either agent as monotherapy. Methods: This phase Ib study of rapid alternation of SU and RE was performed in pts with IM-resistant advanced GIST. A standard 3+3 dosing schema was utilized to determine the recommended phase II dose (RP2D). Pts received continuous treatment with cycles of 3 days of SU followed by 4 days of RE. Plasma samples for pharmacokinetics and ctDNA studies (deep next generation sequencing and ddPCR) were collected at several timepoints. Results: Fourteen pts were enrolled, and 13 received treatment. Median age 64 (range 42-78), 43% female, median prior therapy 4 (range 3-7, all pts had ≥ 3 prior therapies). SU 37.5mg daily 3 days followed by RE 120mg daily 4 days was established as the RP2D. Two dose limiting toxicities (DLTs) occurred at DL2 (asymptomatic G3 hypophosphatemia). Non-DLT G3/4 toxicities were hypertension (1/13 pts) and hand-food syndrome (2/13 pts). 8/13 patients experienced dose modification, delay, or both. No unexpected toxicities were observed. Of the 13 pts with evaluable CT scans, stable disease (SD) was the best response observed in 4 pts by RECIST. Median progression free survival was 8.4 weeks (95% CI of 6.0-15.7). SU and RE did not reach the steady state, although pts with SD had higher median drug concentration than progressing pts. ctDNA studies show that GIST has low ctDNA shedding and remains KIT-driven even at late stages of disease, with a predominance for activation-loop secondary mutations. Conclusions: Rapid alternation of drugs with complementary activity is a well-tolerated treatment strategy. Drug exposure is critical to target effectively specific subpopulations. Although GIST sheds low ctDNA in this heavily pretreated population, identifying KIT muts in plasma is feasible. Clinical trial information: NCT02164240